new feature: JEV Integration. Now there are a current (native) LLM decision model and a JEV decision model

.env.example

@@ -234,3 +234,8 @@ KBDB_MAX_CROSS_REF_HOPS=1
 KBDB_DOCUMENT_TYPE=customer_safe
 KBDB_METADATA_JSON=
 KBDB_MIN_SCORE=
+
+# Framework vocabulary and standard responses: pt-BR | en
+AGENT_LOCALE=pt-BR
+# Optional directory containing complete pt-BR.json and en.json catalogs.
+# AGENT_LOCALES_DIR=./config/locales

.env.jev.example

@@ -0,0 +1,56 @@
+# Copy these settings into the .env loaded by the running backend/gateway.
+# Existing feature flags (ENABLE_LLM_ROUTER, ENABLE_ROUTE_STICKINESS, etc.) still apply.
+# Global default; each individual setting below takes precedence.
+DECISION_PROVIDER=current
+DECISION_FALLBACK_TO_CURRENT=true
+JEV_ENDPOINT=https://api.typesafe.ai/v1/systemone
+AI_GATEWAY_API_KEY=
+JEV_MODEL=jev-latest
+JEV_TIMEOUT_SECONDS=15
+JEV_MIN_CONFIDENCE=0.7
+JEV_DECISION_DIAGNOSTICS=true
+
+# All options: current | jev. No existing behavior changes by default.
+DECISION_ROUTE_CONTINUITY_PROVIDER=current
+DECISION_ROUTER_PROVIDER=current
+DECISION_WORKFLOW_EXPECTED_INPUT_PROVIDER=current
+DECISION_WORKFLOW_REENTRY_PROVIDER=current
+DECISION_TRANSACTION_CONFIRMATION_PROVIDER=current
+DECISION_TRANSACTION_PARAMETER_RELEVANCE_PROVIDER=current
+DECISION_TRANSACTION_INTENT_SHIFT_PROVIDER=current
+DECISION_MULTI_INTENT_PROVIDER=current
+DECISION_TRANSACTION_PARAMETER_RECONCILIATION_PROVIDER=current
+DECISION_GLOBAL_SUPERVISOR_PROVIDER=current
+DECISION_PROCESSING_INTERRUPTION_PROVIDER=current
+
+# Generic LLM guardrail and output supervisor (existing profile names).
+DECISION_GUARDRAIL_GUARDRAIL_PROVIDER=current
+DECISION_GUARDRAIL_GRL_PROVIDER=current
+# Calibrated guardrails, independently by task.
+DECISION_GUARDRAIL_PINJ_PROVIDER=current
+DECISION_GUARDRAIL_COER_PROVIDER=current
+DECISION_GUARDRAIL_TOX_PROVIDER=current
+DECISION_GUARDRAIL_REVPREC_PROVIDER=current
+DECISION_GUARDRAIL_AOFERTA_PROVIDER=current
+DECISION_GUARDRAIL_OOS_PROVIDER=current
+DECISION_GUARDRAIL_FRASEOLOGIA_PROVIDER=current
+DECISION_GUARDRAIL_RAGSEC_PROVIDER=current
+DECISION_GUARDRAIL_DLEX_IN_PROVIDER=current
+DECISION_GUARDRAIL_DLEX_OUT_PROVIDER=current
+
+# Calibrated judges, by task; generic judges, by configured name.
+DECISION_JUDGE_CSI_PROVIDER=current
+DECISION_JUDGE_VCTN_PROVIDER=current
+DECISION_JUDGE_ALUC_PROVIDER=current
+DECISION_JUDGE_RQLT_PROVIDER=current
+DECISION_JUDGE_LLM_JUDGE_PROVIDER=current
+
+# Optional threshold per point, independent from the provider choice.
+# DECISION_ROUTE_CONTINUITY_MIN_CONFIDENCE=0.90
+# DECISION_TRANSACTION_CONFIRMATION_MIN_CONFIDENCE=0.95
+
+# Named calibrated judge overrides its task group (optional).
+# DECISION_JUDGE_GROUNDEDNESS_PROVIDER=jev
+# DECISION_JUDGE_RESPONSE_QUALITY_PROVIDER=current
+# Optional individual guardrail recheck override.
+# DECISION_GUARDRAIL_DLEX_OUT_PROTOCOL_AUTHORIZATION_RECHECK_PROVIDER=jev

.idea/workspace.xml

@@ -4,7 +4,9 @@
     <option name="autoReloadType" value="SELECTIVE" />
   </component>
   <component name="ChangeListManager">
-    <list default="true" id="30a0e1d8-9d7d-469b-b241-f300911cee8a" name="Changes" comment="Ajustes na documentação e remanejamento dos folders" />
+    <list default="true" id="30a0e1d8-9d7d-469b-b241-f300911cee8a" name="Changes" comment="Ajustes na documentação e remanejamento dos folders">
+      <change beforePath="$PROJECT_DIR$/.idea/workspace.xml" beforeDir="false" afterPath="$PROJECT_DIR$/.idea/workspace.xml" afterDir="false" />
+    </list>
     <option name="SHOW_DIALOG" value="false" />
     <option name="HIGHLIGHT_CONFLICTS" value="true" />
     <option name="HIGHLIGHT_NON_ACTIVE_CHANGELIST" value="false" />
@@ -45,34 +47,35 @@
     <option name="hideEmptyMiddlePackages" value="true" />
     <option name="showLibraryContents" value="true" />
   </component>
-  <component name="PropertiesComponent"><![CDATA[{
-  "keyToString": {
-    "ASKED_SHARE_PROJECT_CONFIGURATION_FILES": "true",
-    "ModuleVcsDetector.initialDetectionPerformed": "true",
-    "RunOnceActivity.ShowReadmeOnStart": "true",
-    "RunOnceActivity.TerminalTabsStorage.copyFrom.TerminalArrangementManager": "true",
-    "RunOnceActivity.git.unshallow": "true",
-    "RunOnceActivity.typescript.service.memoryLimit.init": "true",
-    "SHARE_PROJECT_CONFIGURATION_FILES": "true",
-    "Shell Script.agent_frontend.executor": "Run",
-    "Shell Script.agent_template_backend.executor": "Run",
-    "Shell Script.mcp_gateway.executor": "Run",
-    "Shell Script.mcp_server.executor": "Run",
-    "git-widget-placeholder": "master",
-    "kotlin-language-version-configured": "true",
-    "last_opened_file_path": "C:/Asus_Projects/kbdb/agent_framework_oci/templates/agent_template_backend",
-    "node.js.detected.package.eslint": "true",
-    "node.js.detected.package.tslint": "true",
-    "node.js.selected.package.eslint": "(autodetect)",
-    "node.js.selected.package.tslint": "(autodetect)",
-    "nodejs_package_manager_path": "npm",
-    "project.structure.last.edited": "Project",
-    "project.structure.proportion": "0.15",
-    "project.structure.side.proportion": "0.2",
-    "settings.editor.selected.configurable": "advanced.settings",
-    "vue.rearranger.settings.migration": "true"
+  <component name="PropertiesComponent">{
+  &quot;keyToString&quot;: {
+    &quot;ASKED_SHARE_PROJECT_CONFIGURATION_FILES&quot;: &quot;true&quot;,
+    &quot;ModuleVcsDetector.initialDetectionPerformed&quot;: &quot;true&quot;,
+    &quot;RunOnceActivity.ShowReadmeOnStart&quot;: &quot;true&quot;,
+    &quot;RunOnceActivity.TerminalTabsStorage.copyFrom.TerminalArrangementManager&quot;: &quot;true&quot;,
+    &quot;RunOnceActivity.TerminalTabsStorage.copyFrom.TerminalArrangementManager.252&quot;: &quot;true&quot;,
+    &quot;RunOnceActivity.git.unshallow&quot;: &quot;true&quot;,
+    &quot;RunOnceActivity.typescript.service.memoryLimit.init&quot;: &quot;true&quot;,
+    &quot;SHARE_PROJECT_CONFIGURATION_FILES&quot;: &quot;true&quot;,
+    &quot;Shell Script.agent_frontend.executor&quot;: &quot;Run&quot;,
+    &quot;Shell Script.agent_template_backend.executor&quot;: &quot;Run&quot;,
+    &quot;Shell Script.mcp_gateway.executor&quot;: &quot;Run&quot;,
+    &quot;Shell Script.mcp_server.executor&quot;: &quot;Run&quot;,
+    &quot;git-widget-placeholder&quot;: &quot;main&quot;,
+    &quot;kotlin-language-version-configured&quot;: &quot;true&quot;,
+    &quot;last_opened_file_path&quot;: &quot;D:/Dropbox/ORACLE/TIM/FY27/Commits/agent_platform_oci&quot;,
+    &quot;node.js.detected.package.eslint&quot;: &quot;true&quot;,
+    &quot;node.js.detected.package.tslint&quot;: &quot;true&quot;,
+    &quot;node.js.selected.package.eslint&quot;: &quot;(autodetect)&quot;,
+    &quot;node.js.selected.package.tslint&quot;: &quot;(autodetect)&quot;,
+    &quot;nodejs_package_manager_path&quot;: &quot;npm&quot;,
+    &quot;project.structure.last.edited&quot;: &quot;Project&quot;,
+    &quot;project.structure.proportion&quot;: &quot;0.15&quot;,
+    &quot;project.structure.side.proportion&quot;: &quot;0.2&quot;,
+    &quot;settings.editor.selected.configurable&quot;: &quot;advanced.settings&quot;,
+    &quot;vue.rearranger.settings.migration&quot;: &quot;true&quot;
   }
-}]]></component>
+}</component>
   <component name="RecentsManager">
     <key name="CopyFile.RECENT_KEYS">
       <recent name="C:\Asus_Projects\kbdb\agent_framework_oci\templates\agent_template_backend" />
@@ -149,8 +152,8 @@
   <component name="SharedIndexes">
     <attachedChunks>
       <set>
-        <option value="bundled-jdk-9823dce3aa75-fbdcb00ec9e3-intellij.indexing.shared.core-IU-251.29188.36" />
-        <option value="bundled-js-predefined-d6986cc7102b-09060db00ec0-JavaScript-IU-251.29188.36" />
+        <option value="bundled-jdk-30f59d01ecdd-cffe25b9f5b3-intellij.indexing.shared.core-IU-253.28294.334" />
+        <option value="bundled-js-predefined-d6986cc7102b-c7e53b3be11b-JavaScript-IU-253.28294.334" />
       </set>
     </attachedChunks>
   </component>
@@ -174,6 +177,10 @@
       <workItem from="1785409583147" duration="1561000" />
       <workItem from="1785452072692" duration="784000" />
       <workItem from="1785453511270" duration="1006000" />
+      <workItem from="1792665278602" duration="734000" />
+      <workItem from="1793193595355" duration="436000" />
+      <workItem from="1793194079956" duration="5443000" />
+      <workItem from="1793271962068" duration="758000" />
     </task>
     <task id="LOCAL-00001" summary="Ajustes na documentação e remanejamento dos folders">
       <option name="closed" value="true" />

ATUALIZACAO_TESTES.md

@@ -0,0 +1,57 @@
+# Atualização dos testes após mudanças de contrato
+
+Esta revisão mantém os quatro nomes de arquivo e a quantidade de testes. Oito
+testes foram atualizados, sem `skip`, `xfail` ou alterações no runtime para
+acomodar as expectativas antigas. Os nomes de sete funções foram ajustados para
+descrever o comportamento atual. Os comentários nos testes registram o histórico.
+
+## Histórico verificado no Git do pacote original
+
+- `1711575`, 2026-09-07, `bugfix reconciliation extractor`: removeu
+  `field_evidence_views`, a construção de candidatos estruturais e partes da
+  reconciliação anterior.
+- `82636ec`, 2026-09-08, `bugfix: extractor cases 8 and 16`: consolidou a coleta
+  de parâmetros no `single_extractor`, com contexto e pré-validação autoritativa.
+- `36f9781`, 2026-09-18, `new feature: transaction interruption for same intent replacement`:
+  também adicionou sanitização e reavaliação ao DLEX_OUT e passou a usar
+  `sanitized_text` somente quando há um texto substituto.
+
+## O que os testes passam a verificar
+
+| Arquivo | Contrato coberto |
+|---|---|
+| `tests/test_transactional_tool_flow.py` | Após `NEEDS_PARAMETER`, o fluxo aguarda nova informação; a nova validação ocorre no turno seguinte e a operação não executa antes da confirmação. |
+| `tests/test_transaction_parameter_llm_precedence.py` | O reconciliador recebe schema e contexto, aceita o campo resolvido e respeita `preserve` sem sobrescrever o valor conhecido. |
+| `tests/test_transaction_explicit_abandonment_and_structural_candidate.py` | Campo indefinido continua pendente; extração contextual bem-sucedida passa pela validação antes da confirmação; `clear` não inventa um valor nem apaga os demais campos. |
+| `tests/test_dlex_out_expected_protocol_authorization.py` | Protocolo autorizado não exige substituição do texto; credencial é removida e a resposta é reavaliada; outra violação mantém o bloqueio. |
+
+Os simuladores deixaram de depender de frases específicas do prompt, como
+“FOCO PRINCIPAL DA RECONCILIAÇÃO”. Eles devolvem saídas compatíveis com o contrato
+da chamada. A identificação semântica por um modelo real não é comprovada por
+esses testes unitários: os testes verificam como o framework utiliza suas saídas.
+
+## Verificações realizadas
+
+- 160 testes selecionados aprovados no pacote atualizado, incluindo os 41 testes
+  de localização e os quatro arquivos desta revisão.
+- Os 66 testes dos quatro arquivos também foram executados usando a biblioteca
+  original do ZIP: todos aprovados. Assim, a atualização dos testes não depende
+  da alteração de idiomas para passar.
+- Contagem mantida por arquivo: 31, 13, 17 e 5 testes, respectivamente.
+- Nenhuma chamada a OCI, modelos remotos ou ferramentas reais de negócio.
+- Esta revisão não afirma que a suíte completa do repositório foi executada.
+
+## Aplicar ao repositório
+
+O ZIP completo inclui a localização em `pt-BR` e `en` entregue anteriormente e
+esta revisão dos testes. Copie o conteúdo da pasta `agent_platform_oci` para a
+raiz do repositório, preservando a pasta `.git` que já existe no seu checkout.
+
+Se a localização já estiver aplicada, o ZIP complementar de testes contém apenas
+os quatro arquivos atualizados e esta documentação. Não é preciso usar ambos.
+
+Sugestão de mensagem de commit para a revisão dos testes:
+
+```text
+test: align transaction and DLEX_OUT tests with current contracts
+```

JEV_INTEGRATION.md

@@ -0,0 +1,84 @@
+# Seleção individual de decisões: modelo atual ou JEV
+
+Implementação opt-in para o projeto do ZIP de 03/11/2026. O padrão continua sendo `current`; nenhum modelo foi ativado e nenhuma chave foi incluída.
+
+## Configuração
+
+Copie as variáveis de `.env.jev.example` para o `.env` utilizado pelo processo. Se usar contêiner/Kubernetes, passe as variáveis também para o ambiente do processo e guarde a chave no mecanismo de secrets do ambiente. Reinicie o serviço após alterar a configuração. Substitua os arquivos de código e refaça o build/redeploy caso o framework esteja instalado em pacote ou imagem; reiniciar sem atualizar o pacote instalado não aplica mudanças de código.
+
+Exemplo: ativar somente continuidade e intenção, mantendo o restante atual:
+
+```dotenv
+DECISION_PROVIDER=current
+DECISION_ROUTE_CONTINUITY_PROVIDER=jev
+DECISION_ROUTER_PROVIDER=jev
+AI_GATEWAY_API_KEY=<sua chave>
+JEV_MODEL=typesafe-ai/jev
+DECISION_FALLBACK_TO_CURRENT=true
+JEV_MIN_CONFIDENCE=0.7
+DECISION_ROUTE_CONTINUITY_MIN_CONFIDENCE=0.90
+```
+
+A variável individual prevalece sobre `DECISION_PROVIDER`. Uma variável individual ausente herda o padrão global. Como o exemplo define todas como `current`, mudar somente a variável global não altera os pontos explicitamente definidos. Os valores válidos são `current` e `jev`.
+
+A seleção não habilita funcionalidades desativadas: `ENABLE_LLM_ROUTER`, `ENABLE_ROUTE_STICKINESS`, configuração de semantic classifiers e configuração dos guardrails/judges continuam controlando se cada ponto é executado. Para não tornar `current` incompatível, não foram alteradas as regras determinísticas nem os controles de execução.
+
+## Pontos disponíveis
+
+Cada sufixo abaixo corresponde a `DECISION_<SUFIXO>_PROVIDER`:
+
+| Sufixo | Função |
+|---|---|
+| ROUTE_CONTINUITY | CONTINUE, ROUTE, HUMAN_HANDOFF, END_SESSION; inclui sondagem de controles globais que usa o mesmo classificador |
+| ROUTER | Intenção semântica; agente, domínio e ferramentas derivados do catálogo local |
+| WORKFLOW_EXPECTED_INPUT | Opção permitida para retomada de workflow; opção sem correspondência não retoma |
+| WORKFLOW_REENTRY | REENTER ou RESUME |
+| TRANSACTION_CONFIRMATION | Confirmação semântica, usando as opções configuradas; parser explícito permanece |
+| TRANSACTION_PARAMETER_RELEVANCE | RELEVANT ou NOT_RELEVANT |
+| TRANSACTION_INTENT_SHIFT | CONTINUE, REPLACE, SHIFT ou ABANDON e intenção de destino |
+| MULTI_INTENT | Intenção de cada objetivo extraído, com preservação de ordem e trechos literais |
+| TRANSACTION_PARAMETER_RECONCILIATION | resolved, preserve, clear ou unresolved por parâmetro |
+| GLOBAL_SUPERVISOR | Backend permitido no registro do supervisor global |
+| PROCESSING_INTERRUPTION | Regenerar resposta de voz ou repetir |
+| GUARDRAIL_GUARDRAIL / GUARDRAIL_GRL | Guardrails LLM genéricos com os perfis padrão; perfis personalizados usam GUARDRAIL_<PERFIL> |
+| GUARDRAIL_PINJ, COER, TOX, REVPREC, AOFERTA, OOS, FRASEOLOGIA, RAGSEC, DLEX_IN, DLEX_OUT | Cada tarefa calibrada tem sua própria variável GUARDRAIL_<TAREFA> |
+| JUDGE_CSI, JUDGE_VCTN, JUDGE_ALUC, JUDGE_RQLT | Cada tarefa de judge calibrado, como padrão de grupo |
+| JUDGE_<NOME> | Judge genérico, usando seu `name`; o padrão é JUDGE_LLM_JUDGE |
+
+Um judge calibrado também aceita `DECISION_JUDGE_<NOME_CONFIGURADO>_PROVIDER`, que prevalece sobre o grupo da tarefa (por exemplo `DECISION_JUDGE_GROUNDEDNESS_PROVIDER` prevalece sobre `DECISION_JUDGE_ALUC_PROVIDER`). Rechecagens de guardrail aceitam o nome completo do componente, por exemplo `DECISION_GUARDRAIL_DLEX_OUT_PROTOCOL_AUTHORIZATION_RECHECK_PROVIDER`, prevalecendo sobre `DECISION_GUARDRAIL_DLEX_OUT_PROVIDER`.
+
+Caracteres diferentes de letras/números são convertidos em `_`, com nome em maiúsculas. O mesmo ponto pode definir `DECISION_<SUFIXO>_MIN_CONFIDENCE`; esse valor é independente dos limites já existentes no componente. Os limites existentes continuam sendo aplicados depois da integração.
+
+## O que é híbrido
+
+- Multi-intenção: o modelo atual extrai os objetivos e seus trechos, mantendo a ordem. O JEV escolhe a intenção de cada objetivo. Não há geração de trechos pelo JEV; trechos não literais são rejeitados. Isso pode custar mais que uma chamada atual isolada, por ser uma etapa adicional.
+- Reconciliação: o modelo atual extrai candidatos e proveniência. JEV decide o tratamento de cada campo. A decisão não inventa valores e não substitui validação de negócio.
+- Geração final, extração livre, resumos, RAG rewrite/compression, TOXOUT e FALLBACK de texto seguem no provedor atual. Autorização, alçadas, idempotência, pré-validação de negócio e execução seguem nas implementações existentes.
+- Guardrail LLM genérico no JEV avalia se o texto original pode passar sem alteração. Se precisar de sanitização, deve bloquear; o JEV não fornece `sanitized_text` ou instruções textuais de reescrita. Reescritas já separadas no fluxo continuam generativas.
+- Judges JEV não inventam rótulos ou justificativas de prosa: o retorno contém os campos necessários à decisão e motivo técnico fixo. RQLT/CSI usam score 0–10; judge genérico converte para 0–1. Os limites precisam de calibração antes de produção.
+
+## Transporte, falhas e observabilidade
+
+O cliente usa `POST https://ai-gateway.vercel.sh/v1/decisions` com `typesafe-ai/jev`, API key em `Authorization`, timeout de 15 segundos por padrão e perguntas choice/predicate/score. Usa a biblioteca padrão do Python; não adiciona dependências de produção. A chamada bloqueante roda em thread para não bloquear o event loop.
+
+Respostas incompletas, tipos incompatíveis, opções não declaradas, números inválidos, confiança abaixo do limite e erros de transporte acionam o provedor atual uma vez quando `DECISION_FALLBACK_TO_CURRENT=true`. Sem fallback, a exceção chega ao tratamento de falha já existente do componente; isso não redefine automaticamente cada política fail-open/fail-closed. Confiança ausente em choice/score é zero e aciona fallback no limite padrão. Para predicates usa-se a probabilidade do resultado selecionado. Para várias perguntas usa-se o mínimo das confianças.
+
+O custo de uma chamada JEV seguida de fallback inclui ambas as etapas. Respostas recebidas são contabilizadas no repositório de uso e no evento `decision.jev` quando o provedor original oferece esses recursos, mesmo quando a validação rejeita a resposta. Logs registram ponto, modelo, duração, uso e falha/fallback, sem imprimir prompts, chave ou corpo de erro HTTP. Não há geração Langfuse de chat para chamadas JEV; a integração publica evento de decisão.
+
+## Validação
+
+Teste isolado, sem dependências do projeto ou API real:
+
+```sh
+python3 tests/test_jev_transport_standalone.py
+```
+
+Testes adicionais de integração em ambiente com dependências instaladas:
+
+```sh
+PYTHONPATH=libs/agent_framework/src python -m pytest tests/test_jev_decisions.py tests/test_expected_input_semantic_classifier.py tests/unit/test_semantic_route_stickiness.py
+```
+
+O teste isolado cobre contratos, seleção, precedência de configuração, fallback, erros, confiança, transporte e análise sintática dos módulos. Resultado local: 17 testes isolados passaram, incluindo execução dos corpos reais de continuidade, roteamento, reconciliação e judge calibrado com transporte simulado. A suíte completa não foi executada: as dependências do projeto não estavam disponíveis e a tentativa de instalação não conseguiu obtê-las. A suíte completa e a qualidade/latência do modelo real precisam ser verificadas no ambiente de execução. Não foi feita chamada real ao JEV.
+
+Fontes de protocolo consultadas: https://vercel.com/docs/ai-gateway/sdks-and-apis/openai-decisions e https://vercel.com/ai-gateway/models/jev.

LICENSE.txt

@@ -0,0 +1,264 @@
+The Universal Permissive License (UPL), Version 1.0
+This page provides a copy of and also attempts to answer some frequently asked questions about the Universal Permissive License version 1.0. The UPL was first drafted by Oracle in the Spring of 2014, was revised several times after extensive consultation with the Java Community Process Executive Committee and through the Open Source Initiative review process, and was finally approved by the Open Source Initiative in February 2015.
+
+To begin with, this is the final approved text of the Universal Permissive License v 1.0:
+
+Copyright (c) [year] [copyright holders]
+
+The Universal Permissive License (UPL), Version 1.0
+
+Subject to the condition set forth below, permission is hereby granted to any
+person obtaining a copy of this software, associated documentation and/or data
+(collectively the "Software"), free of charge and under any and all copyright
+rights in the Software, and any and all patent rights owned or freely
+licensable by each licensor hereunder covering either (i) the unmodified
+Software as contributed to or provided by such licensor, or (ii) the Larger
+Works (as defined below), to deal in both
+
+(a) the Software, and
+(b) any piece of software and/or hardware listed in the lrgrwrks.txt file if
+one is included with the Software (each a "Larger Work" to which the Software
+is contributed by such licensors),
+
+without restriction, including without limitation the rights to copy, create
+derivative works of, display, perform, and distribute the Software and make,
+use, sell, offer for sale, import, export, have made, and have sold the
+Software and the Larger Work(s), and to sublicense the foregoing rights on
+either these or other terms.
+
+This license is subject to the following condition:
+The above copyright notice and either this complete permission notice or at
+a minimum a reference to the UPL must be included in all copies or
+substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+When referring to the license, you may link to it at https://oss.oracle.com/licenses/upl or you may also find a copy on the Open Source Initiative web site at https://opensource.org/licenses/UPL.
+
+Frequently Asked Questions
+What is the UPL and why did you create it?
+What are the most important features of the UPL?
+Can you point to some concrete problems the UPL solves that you could not have addressed with the MIT, BSD or Apache licenses?
+So why is the absence of an express patent license in the MIT or BSD licenses a problem - isn't the implied license enough?
+Why didn't you include a defensive patent termination provision?
+I heard this licenses our entire patent portfolio to Oracle, is that right?
+How should I apply the UPL to my code?
+What goes in the Larger Works file and what should it look like?
+Can someone else change my Larger Works file and license my patents for other things?
+What is the UPL and why did you create it?
+The UPL is a highly permissive license, including both copyright and patent licenses, which permits use and relicensing under both copyleft and commercial terms, and also facilitates use as a contributor license agreement.
+
+It was originally borne out of a discussion in the Java Community Process around what it would take to permit collaboration on JSR reference implementations in open source forges without the necessity of collaborators signing the JSPA or an Oracle Contributor Agreement. In order to accommodate the broadest possible use case, the license needed to include an express patent license and to be broadly agreed to be GPLv2, GPLv2+Classpath Exception, CDDL, and commercial license compatible, including the ability to cut and paste code between modules, and while it is almost universally agreed among those knowledgeable in FOSS licensing that license proliferation is something to be avoided without good cause, since a license meeting these relatively simple and obvious criteria did not exist, it was decided to draft one.
+
+Drafting started with the MIT license, and after a lot of discussion with other open source lawyers, developers, members of the Java Community Process Executive Committee, and reviewers and board members of the Open Source Initiative, and several rounds of revision, finally arrived at the vetted license text you see here. The license was approved by the OSI as conforming with the Open Source Definition and non-duplicative of existing permissive licenses in February 2015.
+
+Return to FAQ.
+
+What are the most important features of the UPL?
+The most important features and things to understand about the UPL are the following:
+
+Clear patent protection. The UPL is a broad permissive license including both a copyright license and an express patent license, covering at least a version licensed by someone under the license (for example a distributor) and/or a version someone contributed to even if they never distribute the whole. (The reason the latter is needed is discussed below.) By virtue of the unambiguous patent license, the UPL is materially clearer with respect to the rights licensed and likely broader than either the MIT or BSD licenses.
+Clear & simplified relicensing. The UPL expressly permits sublicensing under either the UPL or under other terms, which clearly allows someone to relicense code received under the UPL either on copyleft terms, on proprietary terms, or otherwise, thus permitting maximum flexibility in reuse.
+Reduced overhead in source files. The UPL expressly permits use of the license without including a full copy of the text, which is useful for JavaScript or other cases where minimizing space consumed by licenses is desired - see below for our recommendation of how to apply the license or include proper attribution without the full text.
+It can be used as a contributor agreement. Finally, the UPL may be used as a contributor license agreement licensing both the software itself and also contributor patents for use in one or more "Larger Works." The Larger Works licensed in this fashion are designated by the use of a separate file accompanying the license, akin to the NOTICE file that accompanies the Apache License, Version 2.0. The Larger Works file can be used to control for both contributions to other works (for example, we could specify MySQL in a Larger Works file for a work, which would then license contributor patents for MySQL as well as the contribution), to set patent license scope for specific versions (for example, we could specify the approved reference implementation of JSR-xxx including Maintenance Releases to ensure that all contributors to an RI are licensing both the final version of the RI and qualified updates under the JCP program), or both.
+Return to FAQ.
+
+Can you point to some concrete problems the UPL solves that you could not have addressed with the MIT, BSD or Apache licenses?
+The most important problem the UPL solves is the inclusion of an express patent license in a license that is both broadly agreed to be uniformly copyleft compatible and also permits the software to be freely used in commercially licensed software (i.e., without concerns about reciprocal license obligations).
+
+The MIT and BSD licenses do not include express patent licenses, and while many argue that there is an implied patent license, there is nontrivial debate about the existence & scope of any such implied license. This is addressed in some more detail below. The Apache license, on the other hand, is argued by some not to be GPLv2 compatible, which makes it tricky to use in connection with GPLv2 licensed code - the GPLv2 is still by far the most popular open source license, and unambiguous, undisputed GPL compatibility was a crucial qualifier for any license considered for use in the JCP. (Even the BSD license has been argued not to be GPL compatible - while this need not be addressed here, and it is not the conclusion of the FSF or most others, by virtue of expressly permitting relicensing on other terms, the UPL is definitively and fully GPLv2 compatible.)
+
+Another problem solved with the UPL, which seems like a small thing but adds compliance and clarity, is that licensors and distributors are expressly permitted to reference the UPL rather than including a complete copy of the text in each item. What is the license solving for here? The MIT and BSD licenses expressly require that you include a complete copy of the license in each piece of source code. This means that if you have 50 different JavaScript files being downloaded from a page, each one should be giving you a complete copy of the license - you may be using as much bandwidth for licenses as you do for the actual code! And if you do not include a complete copy of the license, as some folks do not, merely referring to the license by URL or name, then you are not really applying the license in the intended manner and are in fact suggesting that downstream recipients not strictly follow those terms either.
+
+Finally, the inclusion of the Larger Works concept facilitates use as a contributor license agreement (CLA), with contributors granting a patent license to the works to which they're contributing, and therefore creating a common & safe platform for collaboration where no one is going to assert infringement by the same project in which they're an active participant.
+
+Return to FAQ.
+
+So why is the absence of an express patent license in the MIT or BSD licenses a problem - isn't the implied license enough?
+Some authorities think it is enough, but not everyone agrees on this point. To begin with, it has been suggested that in some jurisdictions there may be no implied license at all, or only a license to people to whom a licensor directly distributed the software.
+
+But even if one were to take the position that a patent license covering any contribution is implied, and further that it covers the entirety of a work you distribute, in our world of cloud offerings, many contributors may never distribute the entirety of the work at all. This may reduce the likelihood that a court would find them to have granted an implied license for the whole of the work.
+
+The UPL attempts to solve these problems by having each party that either contributes to the project or distributes it under the UPL expressly license all patents that they own or can freely license that cover the work - thus, e.g., licensing the whole even from contributors who never distribute the whole.
+
+Return to FAQ.
+
+Why isn't there a defensive patent termination provision?
+A defensive patent termination provision may sound like a great idea at first, especially for a pure open source project, but does not really work for contributions to software that is also licensed on commercial terms.
+
+Why? A company that takes on commercial obligations with respect to a piece of code needs to retain the ability to keep supporting and shipping that code. A defensive patent termination provision allows someone who makes a contribution and then breaches the project license to the project to then potentially effectively block infringement suits against them, even when they have willfully and flagrantly disregarded community norms and IP rights in the project. Such a provision is therefore in reality as much of a sword as it is a shield.
+
+Say, for example, that a fictional company WidgetCo were to make a contribution to fictional company GadgetCo's product GadgetX under a license with a patent termination provision. Now, WidgetCo decides to violate GadgetCo's terms and distribute GadgetX without a license. GadgetCo's options for stopping the violation would ordinarily include both copyright and also patent infringement claims for any patents covering GadgetX. If, on the other hand, WidgetCo has made patent covered contributions under an agreement with a so-called defensive termination clause, GadgetCo's ability to stop WidgetCo from violating the license may be significantly compromised because it can no longer bring patent claims without risking losing its own license to the patents on the contributions, which therefore might require it to stop shipping GadgetX immediately and stop supporting versions that customers are relying on.
+
+As a commercial distributor of software, GadgetCo's customers rely on it to have the rights it needs to be able to continue to deliver and support the software, and defensive patent termination rights in contributions are inconsistent with those obligations. Defensive patent termination only helps someone who may be subject to infringement litigation by someone who is using the code - it won't stop patent trolls, who aren't normally using the code to begin with. So who does it stop, in reality? Someone who is using the code but has IP rights that someone else using the code is infringing. Now, it may be reasonable to tell someone who has no relationship to a community that they cannot simultaneously expect to be able to use the code and also sue for its use - thus making such provisions at least not entirely crazy for a pure FOSS project that no one is signing up to develop or support on a commercial basis, and which is rarely if ever going to need to enforce its rights anyway. But for anyone who both actually expects to need to enforce its license and also wants to take contributions, this makes no sense - you need to be able to enforce your rights in the project without needing to stop shipping or using it, and accepting contributions under terms with a patent termination provision does not satisfy this basic need.
+
+Return to FAQ.
+
+I heard this licenses our entire patent portfolio to Oracle, is that right?
+No, it is not, this is simply untrue except in odd and contrived circumstances that would apply to other FOSS licenses as well. If your entire patent portfolio happens to read on a UPL licensed project that you have distributed on those terms or contributed to, or its designated Larger Works, then yes, it would all be licensed for use in that Software and those specified Larger Works, not just to Oracle but to all licensees under the UPL, just as it would all be licensed for a project under the Apache Software License or Eclipse Public License if your entire portfolio read on a project under the ASL or EPL. Patents that do not read on the project as you contributed to or distributed it or the specified Larger Works are not licensed, there is no magic here.
+
+The patent scope is not really all that complicated. It covers your contributions, the project you contributed to in the state it existed when you contributed, and anything specified in the Larger Works file (which may be another application, it may be other or later versions, it may be a piece of hardware, or any combination of those things - or, just as importantly, it may be none of those things, since many projects may have no Larger Works file at all, in which case the patent license is little different than the express patent license in the Apache license, only without the so-called defensive termination provision).
+
+Return to FAQ.
+
+How should I apply the UPL to my code?
+It is best to insert a copyright notice and a copy of the license at the top of each source file, and if there is a package of numerous files, to also insert a LICENSE file including a copy of the UPL in the root of the distribution. Here is an example:
+
+/*
+** MyWidget version 1.0.
+**
+** Copyright (c) 2015 WidgetCo, Inc.
+**
+** The Universal Permissive License (UPL), Version 1.0
+**
+** Subject to the condition set forth below, permission is hereby granted to any
+** person obtaining a copy of this software, associated documentation and/or data
+** (collectively the "Software"), free of charge and under any and all copyright
+** rights in the Software, and any and all patent rights owned or freely
+** licensable by each licensor hereunder covering either (i) the unmodified
+** Software as contributed to or provided by such licensor, or (ii) the Larger
+** Works (as defined below), to deal in both
+** 
+** (a) the Software, and
+** (b) any piece of software and/or hardware listed in the lrgrwrks.txt file if
+** one is included with the Software (each a "Larger Work" to which the Software
+** is contributed by such licensors),
+** 
+** without restriction, including without limitation the rights to copy, create
+** derivative works of, display, perform, and distribute the Software and make,
+** use, sell, offer for sale, import, export, have made, and have sold the
+** Software and the Larger Work(s), and to sublicense the foregoing rights on
+** either these or other terms.
+** 
+** This license is subject to the following condition:
+** The above copyright notice and either this complete permission notice or at
+** a minimum a reference to the UPL must be included in all copies or
+** substantial portions of the Software.
+** 
+** THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+** IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+** FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+** AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+** LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+** OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+** SOFTWARE.
+*/
+If using the UPL for JavaScript or other circumstances in which space is at a premium, you may use a reference to the license rather than a full copy (though a full copy is always preferred if there's room and you won't be transmitting a bunch of copies to the same user). If you want to use just a short reference, you should still insert a copyright notice, and along with it, a statement that the file is licensed under the UPL, and may link to it at https://oss.oracle.com/licenses/upl or on the Open Source Initiative web site at https://opensource.org/licenses/UPL.
+
+Here is an example header with a short reference for use in source code:
+
+/*
+** MyWidget version 1.0.
+**
+** Copyright (c) 2015 WidgetCo, Inc.
+** Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/
+*/
+Return to FAQ.
+
+What goes in the Larger Works file and what should it look like?
+To begin with, we should note that the Larger Works concept will be irrelevant to many people and to be clear, there is no need to license any Larger Works at all in order to use the UPL on your code - this is still advantageous versus other permissive licenses because licensees can count on a patent license of unambiguous scope from each contributor, can freely relicense the code as necessary for their needs, and can abbreviate the license notice if appropriate.
+
+The purpose of the Larger Works addition to the license is to allow you to use the UPL as a contributor license agreement, commonly referred to as a CLA, licensing patents for a specified larger work or set of works with which contributions may be used. If you desire to use the UPL as a CLA, in addition to including the UPL and/or references to it in your source, you should insert a lrgrwrks.txt file in the root of the distribution specifying the licensed Larger Works. You will name the licensed project/product there, and if appropriate, specify what versions are covered. In theory you could use a single word naming a product, which would be effective, but some slightly prettier examples are as follows.
+
+Here is what we are hoping to use in connection with JCP Reference Implementations:
+
+/* Larger Works File for JSR-xxx
+** 
+** Licensed Larger Works: The Reference Implementation for JSR-xxx Including Maintenance Releases
+** as defined in the JCP Procedures documentation.
+*/   
+So what is the intent here? We are not expanding the scope of the licensed project beyond the RI, however, we want to make sure that the RI is fully covered such that an early contributor who stops adding new code still licenses the RI as a whole, including Maintenance Releases down the line.
+
+A somewhat broader example is as follows:
+
+/* Larger Works File for MyWidget
+** 
+** Larger Works: MyApp
+*/
+Here, someone has set up the Larger Works file to act as a CLA for the application MyApp. The licensors of MyWidget are also licensing their patents for use in MyApp, and because there is no version restriction, this is presumably true for future versions of MyApp as well as the current version. If they wanted to license only the current version, they could have said:
+
+/* Larger Works File for MyWidget
+** 
+** Larger Works: MyApp 1.0
+*/
+Or if Company A wanted to use the UPL with a Larger Works file as a CLA for all of their products, they could use one of the following:
+
+/* Larger Works File for Contributions to Company A Projects
+** 
+** Larger Works: All Software Authored or Distributed by Company A
+*/
+or
+
+/* Larger Works File for Contributions to Company A Projects
+** 
+** Larger Works: All Company A Products and Projects
+*/
+If the Apache Software Foundation wanted to use the UPL as a CLA, they could use the following:
+
+/* Larger Works File for Contributions to ASF Projects
+** 
+** Larger Works: All Apache Software Foundation Projects
+*/
+or even more broadly, to include dependencies that are not ASF projects:
+
+/* Larger Works File for Contributions to ASF projects
+**  
+** Larger Works: All Apache Software Foundation projects and their dependencies
+*/
+or they could make it project specific, for single or multiple Projects:
+
+/* Larger Works File for Contributions to Apache Ant & Maven
+** 
+** Larger Works: Apache Ant and Apache Maven and their subprojects
+*/
+or project and version specific:
+
+/* Larger Works File for Contributions to Cassandra
+** 
+** Larger Works: Apache Cassandra 2.0.15 and updates other than major version updates
+*/
+or subject area specific:
+
+/* Larger Works File for Contributions to Build Tools
+** 
+** Larger Works: All Apache Software Foundation projects pertaining to software builds and related processes, including without
+** limitation Ant, Ant Libraries, Ivy, IvyDE, EasyAnt, Maven and other software for controlling build processes tools
+*/
+As you can see, the intent of the Larger Works file is to permit nearly infinite flexibility in what is being licensed beyond the Software itself, as every company and FOSS project has different needs. Note that while a comment format has been used for these examples, since this is not executable code, this isn't strictly necessary, it was done as a matter of habit and in case someone imports one of the files into some executable code.
+
+Return to FAQ.
+
+Can someone else change my Larger Works file and license my patents for other things?
+To begin with, a party who alters the Larger Works file cannot alter the scope of the license someone else granted, any more than they could alter the license text of the UPL, the GPL, or any other license and assert that the original licensors are bound by their changes. So no one can offer a broader license to your patents than the one you declared. The only power they have is to sublicense downstream recipients on different terms under whatever IP rights they have.
+
+And on the flip side, what happens if someone simply removes the Larger Works file from a code base and then redistributes without it? Then they are sublicensing the code to a downstream recipient on other terms, without any Larger Works being licensed under their own patents, though the license from the original licensors would still survive on those terms as to people who receive the Software from other sources.
+
+Similarly, if someone alters the Larger Works file and then distributes the code to new recipients themselves under the UPL, then that new distributor is granting a license of a different scope than the one granted by the original authors, which they are permitted to do, just as they would be permitted to grant their own license to the code under the GPL or a commercial license. However, the polite thing to do in this case would be to address the altered scope of their license in the Larger Works file while leaving intact the Larger Works designation from prior licensors. Let's say, for example, we start off with a Larger Works file that looks like this, covering all of Java:
+
+/* Larger Works File for Contributions to Project X
+** 
+** Larger Works: All Oracle Java implementations including without limitation Oracle Java EE, Java SE, Java ME, and OpenJDK
+*/
+This would grant a license to all of Java from each contributor on those terms. Now let's say Company A only wants to license a particular JSR, not all of Java. They would ideally add separate license and Larger Works files that travel with their contributions calling out only that JSR, most effectively by placing them in a separate directly clearly demarcating what is under that license. (Separating these things out is always a trick when you try to commingle source in the same file, but if you keep files separate, this is not generally difficult.)
+
+In the alternative, if the contributions were not easily separable from the larger project in another directory, and for some reason they could not add a separate directory just for the license either, and therefore absolutely needed a way to combine their own, distinct Larger Works designation with that from others, they could in theory also add their own declaration to the existing Larger Works file without purporting to change the license from others as follows:
+
+/* Larger Works File for Contributions to Project X
+** 
+** Larger Works: All Oracle Java implementations including without limitation Oracle Java EE, Java SE, Java ME, and OpenJDK
+**
+** Exception: Company A licenses only the following Larger Works: The Reference Implementation for JSR-xxx including
+** Maintenance Releases as defined in the JCP Procedures documentation
+*/
+This would therefore clearly inform a recipient that only the RI was licensed under Company A's patents, but leave intact the reference to the broader license from other contributors.
+
+There are certainly other possibilities for accommodating multiple Larger Works scopes as well, about which your counsel can advise you.
+
+Return to FAQ.
+
+Return to top.

LOCALIZACAO.md

@@ -0,0 +1,133 @@
+# Termos e frases configuráveis — pt-BR e en
+
+Esta distribuição inclui dois catálogos completos, em JSON UTF-8:
+
+- `libs/agent_framework/src/agent_framework/locales/pt-BR.json`
+- `libs/agent_framework/src/agent_framework/locales/en.json`
+
+Cada arquivo reúne termos reconhecidos, mensagens padronizadas, instruções de
+recuperação de guardrails e padrões de extração. JSON permite editar o catálogo
+sem uma dependência adicional no carregador. Os dois arquivos usam as mesmas chaves.
+
+## Selecionar o idioma
+
+No `.env` do backend:
+
+```dotenv
+AGENT_LOCALE=en
+```
+
+Para português, use `AGENT_LOCALE=pt-BR` ou omita a variável. Também são aceitos
+`pt`, `pt_BR`, `en-US` e `en-GB`. Uma opção inexistente gera erro; não há troca
+silenciosa para português quando o idioma solicitado não está disponível.
+
+O idioma é o padrão do backend. Os prompts e as configurações de negócio dos
+agentes continuam sendo definidos pelo projeto; use prompts e descrições no
+idioma desejado. Esta alteração não traduz automaticamente conteúdos de RAG,
+respostas dos servidores MCP, mensagens declaradas por um workflow, arquivos de
+exemplo, regras de negócio ou textos personalizados do aplicativo/frontend.
+Prompts internos de classificação podem continuar em português: as decisões são
+estruturadas. As instruções de recuperação que impunham português foram separadas
+nos catálogos, incluindo sua variante em inglês.
+
+## Personalizar sem alterar a biblioteca
+
+Copie os catálogos para um diretório da aplicação e configure:
+
+```dotenv
+AGENT_LOCALE=en
+AGENT_LOCALES_DIR=./config/locales
+```
+
+Nesse exemplo, o arquivo usado será `./config/locales/en.json`. O catálogo é
+completo, e não um arquivo de alterações parciais. Preserve as chaves e os campos
+como `{action}`, `{target}`, `{options}` e `{subject}` ao traduzir as frases.
+Não é necessário manter a mesma quantidade de termos nas listas.
+
+Reinicie o backend após editar os arquivos: o carregador mantém os catálogos em
+cache. Na inicialização, ele verifica a estrutura, os campos das mensagens, a
+sintaxe dos padrões e conflitos entre confirmação e recusa.
+
+As principais seções são:
+
+| Seção | Conteúdo |
+|---|---|
+| `terms.confirmation` | Confirmações e recusas explícitas de transações |
+| `terms.workflow` | Respostas de compatibilidade para checkpoints antigos |
+| `terms.multi_intent` | Conjunções, verbos e prefixos de confirmação |
+| `terms.global_router` | Termos de continuidade e encerramento |
+| `terms.groundedness` | Indícios de especificidade e incerteza |
+| `terms.mock` | Termos dos classificadores locais usados em testes/desenvolvimento |
+| `messages` | Frases de confirmação, nova tentativa, esclarecimento e fallback |
+| `prompts.fallback` | Instruções de reescrita e regeneração por guardrail |
+| `prompts.judge_rewrite` | Instruções de recuperação dos judges |
+| `memory` | Padrões para extração de memória de longo prazo |
+| `protocol` | Dígitos, letras e nomes usados na vocalização de protocolos |
+| `safety` | Padrões determinísticos de segurança e escopo |
+| `legacy_extraction` | Padrões do extrator legado de pedido/motivo |
+
+## Decisões internas e contratos de workflow
+
+Os códigos internos permanecem estáveis: `confirm`, `reject`, `SIM`, `NAO`,
+`OUTRO` e nomes de estados não devem ser traduzidos em arquivos existentes.
+O catálogo trata as palavras reconhecidas e os textos apresentados.
+
+Um workflow que declara `expected_input`, `reprompt` ou mensagens próprias
+continua tendo precedência. Para aceitar `YES` e `NO` em um contrato enumerado,
+configure essas opções no workflow e ajuste suas transições, ou mantenha os
+códigos atuais e use seu classificador semântico para mapear as respostas.
+O dicionário global não substitui silenciosamente o contrato de um fluxo.
+
+As listas de confirmação transacional usam correspondência da frase completa,
+ignorando maiúsculas, espaços repetidos e pontuação final simples. Não inclua
+frases ambíguas como “entendi”/“understood” nessas listas: compreender uma
+explicação não equivale a autorizar uma operação. O fallback semântico existente
+continua disponível para respostas que não constam do catálogo.
+
+Os padrões de segurança mantêm cobertura em português e inglês nos dois arquivos.
+Nomes de concorrentes presentes em `safety` preservam a configuração original de
+telecom; adapte essas regras quando mudar o domínio da aplicação.
+
+## Vários idiomas no mesmo processo
+
+Para aplicações que selecionam o idioma por agente ou atendimento, envolva a
+execução completa (roteamento, runtime e guardrails) no contexto correspondente:
+
+```python
+from agent_framework.localization import use_locale
+
+with use_locale("en"):
+    # Execute aqui o roteamento, o agente e seus guardrails.
+    result = await handle_turn(...)
+```
+
+`handle_turn` representa o ponto de entrada da sua aplicação; não é uma nova API
+do framework. O contexto usa `ContextVar`, é isolado entre tarefas assíncronas e
+restaura o idioma anterior ao sair. Não altere `os.environ` a cada atendimento.
+O idioma da conversa não é detectado automaticamente.
+
+## Validação desta entrega
+
+- 41 testes novos de localização aprovados.
+- 160 testes selecionados aprovados após a atualização de oito testes antigos.
+- As oito expectativas antigas estavam desalinhadas com mudanças de contrato
+  anteriores à localização. Histórico, escopo e validação estão em
+  `ATUALIZACAO_TESTES.md`. Os 66 testes dos quatro arquivos atualizados também
+  passaram com a biblioteca original do ZIP.
+- As verificações incluem execução/cancelamento real do runtime com ferramentas
+  simuladas, confirmação determinística no roteador sem LLM, múltiplas intenções,
+  retomada de workflows, memória, protocolos, segurança, concorrência e catálogo
+  personalizado.
+- Não foram feitas chamadas a modelos remotos, OCI ou ferramentas reais de negócio.
+
+Os testes novos estão em `tests/test_locale_catalogs.py`. Para executá-los em um
+ambiente com as dependências do projeto e pytest/pytest-asyncio:
+
+```bash
+PYTHONPATH=libs/agent_framework/src pytest -q tests/test_locale_catalogs.py
+```
+
+O pacote de entrega conserva os arquivos do projeto e remove metadados do macOS,
+cache de bytecode, builds antigos e diretórios de Git/IDE. As fontes corrigidas
+ficam em `libs/agent_framework/src`. As variantes de demonstração em
+`Tuning-Performance` mantêm suas configurações de domínio originais.

OBSERVABILITY_FIX_NOTES.md

@@ -0,0 +1,17 @@
+# Observability severity patch
+
+- Central `severity_policy.py` classifies explicit MCP failures, guardrail blocks, judge rejections, RAG failures and memory/cache errors.
+- `Telemetry.event()` accepts `level` and `status_message` and preserves WARNING/ERROR events as individual observations in compact mode.
+- `Telemetry.span()` handle supports `set_status(level, message)`; failed MCP results mark their span ERROR without raising exceptions.
+- Long-term-memory read/write failures emit telemetry while preserving existing fallback behavior.
+- Guardrail blocked and negative judge evaluation emit WARNING when these helper methods are invoked.
+- Normal cache misses and empty RAG results are not automatically failures.
+
+## Scope / limitations
+- Existing call sites must invoke `GuardrailTelemetry.blocked` for guardrail blocking events; this patch does not rewrite every pipeline call site.
+- Redis cache currently has no injected Telemetry instance, so its existing logger errors are not yet Langfuse observations. A separate dependency-injection change is required to cover that path safely.
+- RAG database connection errors that are caught by external KBDB integrations without emitting a telemetry event cannot be classified automatically.
+- No live Langfuse ingestion test was possible in this environment. Verify both compact and verbose modes against your server.
+
+## Validation
+`PYTHONPATH=libs/agent_framework/src python -m pytest -q tests/test_observability_severity_policy.py`

SECURITY.md

@@ -0,0 +1,38 @@
+# Reporting security vulnerabilities
+
+Oracle values the independent security research community and believes that
+responsible disclosure of security vulnerabilities helps us ensure the security
+and privacy of all our users.
+
+Please do NOT raise a GitHub Issue to report a security vulnerability. If you
+believe you have found a security vulnerability, please submit a report to
+[secalert_us@oracle.com][1] preferably with a proof of concept. Please review
+some additional information on [how to report security vulnerabilities to Oracle][2].
+We encourage people who contact Oracle Security to use email encryption using
+[our encryption key][3].
+
+We ask that you do not use other channels or contact the project maintainers
+directly.
+
+Non-vulnerability related security issues including ideas for new or improved
+security features are welcome on GitHub Issues.
+
+## Security updates, alerts and bulletins
+
+Security updates will be released on a regular cadence. Many of our projects
+will typically release security fixes in conjunction with the
+Oracle Critical Patch Update program. Additional
+information, including past advisories, is available on our [security alerts][4]
+page.
+
+## Security-related information
+
+We will provide security related information such as a threat model, considerations
+for secure use, or any known security issues in our documentation. Please note
+that labs and sample code are intended to demonstrate a concept and may not be
+sufficiently hardened for production use.
+
+[1]: mailto:secalert_us@oracle.com
+[2]: https://www.oracle.com/corporate/security-practices/assurance/vulnerability/reporting.html
+[3]: https://www.oracle.com/security-alerts/encryptionkey.html
+[4]: https://www.oracle.com/security-alerts/

apps/agent_frontend_a2a/Dockerfile

@@ -0,0 +1,3 @@
+FROM nginx:stable-alpine
+COPY index.html styles.css app.js a2a-client.js /usr/share/nginx/html/
+EXPOSE 80

apps/agent_frontend_a2a/README.md

@@ -0,0 +1,133 @@
+# Frontend A2A para o Agent Framework OCI
+
+Interface estática, sem build e sem dependências JavaScript de produção. Usa os
+endpoints A2A que já existem no pacote: pode falar com o template de backend
+ou com o Agent Gateway. O frontend nativo em `apps/agent_frontend` é preservado.
+
+## Experimentar sem OCI
+
+Na raiz do framework:
+
+```sh
+python -m pip install -e 'libs/agent_framework[a2a]' uvicorn
+python -m uvicorn examples.a2a_frontend_demo:app --host 127.0.0.1 --port 8000
+```
+
+Abra **http://localhost:8000** e clique em **Conectar ao serviço**. O exemplo
+serve a tela e a API na mesma origem, sem precisar configurar CORS.
+
+A demonstração usa o **adaptador A2A e o SDK reais**, com a lógica de negócio
+simulada. Digite uma mensagem e acompanhe a tarefa. Comandos de demonstração:
+
+- `/confirmar`: aguarda sua resposta e retoma a mesma Task.
+- `/lento`: permite testar o botão de cancelamento.
+- `/erro`: demonstra uma Task que termina em falha.
+
+Se a porta 8000 estiver ocupada, altere a porta ao iniciar e informe o novo
+endereço no campo **Endereço do serviço**, ou use `?service=http://localhost:NOVA_PORTA`.
+
+## Usar com seu agente ou Gateway
+
+Habilite A2A no serviço, conforme `docs/A2A_ASSINCRONO.md`, e sirva este diretório:
+
+```sh
+python -m http.server 5173 --directory apps/agent_frontend_a2a
+```
+
+Abra **http://localhost:5173**. Não abra `index.html` como arquivo local, pois a
+interface usa módulos JavaScript e requisições HTTP.
+
+| Uso | Endereço do serviço | Destino na interface | Seleção carregada do Agent Card |
+| --- | --- | --- | --- |
+| Template de agente | `http://localhost:8000` | Agente | Perfis de `agents.yaml` |
+| Agent Gateway | `http://localhost:8010` | Gateway | Backends do catálogo |
+
+Use **Seleção automática** para deixar a decisão com o serviço. No Gateway,
+um backend selecionado é enviado como `message.metadata.backend_id`. Um perfil
+opcional dentro dele pode ser informado em **Perfil do agente no backend**.
+O adaptador A2A encaminha a seleção ao roteador existente; os contratos nativos
+não são alterados.
+
+O serviço deve permitir a origem da interface em `CORS_ORIGINS`. O valor padrão
+do pacote já permite `http://localhost:5173`. Se mudar o host ou a porta,
+configure a origem exata, por exemplo:
+
+```dotenv
+CORS_ORIGINS=http://localhost:5173,http://127.0.0.1:5173
+ENABLE_A2A=true
+A2A_TASK_MODE=true
+```
+
+Para publicar com Docker:
+
+```sh
+docker build -t agent-frontend-a2a apps/agent_frontend_a2a
+docker run --rm -p 5173:80 agent-frontend-a2a
+```
+
+O navegador precisa alcançar o endereço informado no formulário. Para uso
+externo, informe a URL pública do backend/Gateway, não o nome interno do
+container. Uma interface HTTPS deve acessar um serviço HTTPS.
+
+## O que a interface oferece
+
+- Descoberta do Agent Card e escolha de versão automática ou explícita (1.0/0.3).
+- Envio de mensagem com retorno imediato de Task.
+- Acompanhamento por POST SSE com `SubscribeToTask`/`tasks/resubscribe`.
+- Recuperação por `GetTask`/`tasks/get` quando o fluxo termina ou cai.
+- Consulta periódica como alternativa ao streaming.
+- Cancelamento solicitado ao serviço, incluindo tarefas aguardando input.
+- Continuação de tarefas em input-required/auth-required.
+- Exibição de respostas diretas e dos artefatos de Tasks, com suporte aos
+  resultados estruturados de múltiplos agentes do framework.
+- Contexto de negócio opcional (cliente e contrato/pedido).
+- Identificadores da conversa/tarefa preservados em `sessionStorage` para
+  retomar após recarregar a mesma aba. O conteúdo do chat não é persistido.
+- Nova conversa independente, sem reaproveitar uma Task concluída.
+
+O frontend usa `fetch` para SSE, permitindo enviar os mesmos cabeçalhos de
+autenticação nas consultas e no acompanhamento. Em **Contexto e autenticação**,
+informe o nome e o valor completo do cabeçalho exigido pelo serviço, por
+exemplo `Authorization` / `Bearer ...`, ou `x-api-key` / valor da chave.
+Credenciais ficam somente na memória e precisam ser informadas novamente após
+recarregar. Não são gravadas em `sessionStorage` nem colocadas na URL.
+
+Nenhuma operação de envio é repetida automaticamente. Se a resposta do envio
+não chegar, a tela informa a incerteza, pois o serviço pode ter aceitado a
+mensagem. O botão **Retomar** consulta a mesma Task, quando seu ID é conhecido.
+Trocar conexão, tenant, perfil ou backend requer reconectar e inicia outra
+conversa quando a identidade da configuração mudou.
+
+## Limites herdados do serviço
+
+O frontend não transforma o armazenamento em memória em armazenamento durável.
+Uma Task perdida ao reiniciar o serviço não pode ser recuperada pela tela.
+O cancelamento local não desfaz efeitos externos já confirmados e não garante
+cancelar uma operação já aceita por um agente remoto.
+
+`auth-required` indica que a tarefa aguarda autorização, mas esta interface não
+implementa login OAuth nem aprovação externa automática. Realize o fluxo exigido
+pelo agente, atualize a credencial se necessário e retome a tarefa.
+
+Não há upload de arquivos, push notifications, gRPC ou emissão de tokens do LLM.
+A seleção de versão 0.3 é explícita quando o serviço publica um Agent Card 1.0
+mas também aceita o formato 0.3, como ocorre no adaptador do framework.
+
+## Testes
+
+Sem bibliotecas JavaScript adicionais, com Node.js 22 ou superior:
+
+```sh
+cd apps/agent_frontend_a2a
+node --test tests/client.test.js
+```
+
+Com a demonstração em execução:
+
+```sh
+A2A_DEMO_URL=http://localhost:8000 node --test tests/*.test.js
+```
+
+Os testes de integração exercitam o cliente JavaScript contra o SDK real do
+framework: A2A 1.0 e 0.3, envio, streaming, consulta, confirmação, cancelamento
+e falha. Os testes de interface também foram realizados em navegador local.

apps/agent_frontend_a2a/a2a-client.js

@@ -0,0 +1,167 @@
+export const uid = () => globalThis.crypto.randomUUID();
+export const stateName = (state = '') => state.replace(/^TASK_STATE_/, '').toLowerCase().replaceAll('_', '-');
+export const isStopped = state => ['completed', 'failed', 'canceled', 'rejected', 'input-required', 'auth-required'].includes(stateName(state));
+export const needsInput = state => ['input-required', 'auth-required'].includes(stateName(state));
+
+export class RpcError extends Error {
+  constructor(error) { super(error.message || 'O serviço retornou um erro.'); this.name = 'RpcError'; this.code = error.code; }
+}
+export function serviceURLs(value) {
+  const url = new URL(value);
+  if (!['http:', 'https:'].includes(url.protocol) || url.username || url.password || url.search || url.hash) throw new Error('Informe uma URL HTTP ou HTTPS sem credenciais, parâmetros ou fragmentos.');
+  const base = url.href.replace(/\/$/, '').replace(/\/a2a$/, '');
+  return { base, rpc: `${base}/a2a`, card: `${base}/.well-known/agent-card.json` };
+}
+export function normalizeResult(result) {
+  if (!result || typeof result !== 'object') throw new Error('Resposta A2A inválida.');
+  for (const [key, type] of [['task','task'], ['message','message'], ['statusUpdate','status'], ['artifactUpdate','artifact']]) {
+    if (result[key]) return { type, value: result[key] };
+  }
+  const kinds = { task: 'task', message: 'message', 'status-update': 'status', 'artifact-update': 'artifact' };
+  if (kinds[result.kind]) return { type: kinds[result.kind], value: result };
+  throw new Error('O serviço não retornou uma mensagem, tarefa ou evento A2A reconhecido.');
+}
+export function selectVersion(card, preference = 'auto') {
+  if (preference !== 'auto') return preference;
+  const versions = (card.supportedInterfaces || []).filter(i => i.protocolBinding === 'JSONRPC').map(i => i.protocolVersion);
+  if (versions.some(v => /^1\./.test(v))) return '1.0';
+  if (/^0\.3/.test(card.protocolVersion || '') || versions.some(v => /^0\.3/.test(v))) return '0.3';
+  throw new Error('O Agent Card não anuncia A2A 1.0 ou 0.3 por JSON-RPC.');
+}
+
+// SSE over POST/fetch supports authentication headers and both A2A versions.
+export async function* parseSSE(body) {
+  if (!body) throw new Error('O serviço não abriu o fluxo de eventos.');
+  const reader = body.getReader();
+  const decoder = new TextDecoder();
+  let buffer = '', data = [], size = 0;
+  try {
+    while (true) {
+      const { value, done } = await reader.read();
+      buffer += done ? decoder.decode() : decoder.decode(value, { stream: true });
+      while (true) {
+        const end = buffer.search(/[\r\n]/);
+        if (end < 0 || (!done && buffer[end] === '\r' && end === buffer.length - 1)) break;
+        const line = buffer.slice(0, end);
+        buffer = buffer.slice(end + (buffer.slice(end, end + 2) === '\r\n' ? 2 : 1));
+        if (!line) {
+          if (data.length) yield data.join('\n');
+          data = []; size = 0;
+        } else if (line.startsWith('data:')) {
+          const entry = line.slice(5).replace(/^ /, '');
+          data.push(entry); size += entry.length;
+        }
+        if (size > 4 * 1024 * 1024) throw new Error('Evento recebido excedeu o limite de tamanho.');
+      }
+      if (buffer.length > 4 * 1024 * 1024) throw new Error('Evento recebido excedeu o limite de tamanho.');
+      if (done) break; // A partial frame is not an event; the caller can recover with GetTask.
+    }
+  } finally {
+    await reader.cancel().catch(() => {});
+    reader.releaseLock();
+  }
+}
+
+export class A2AClient {
+  constructor({ url, version = '1.0', headers = {}, fetchImpl = globalThis.fetch }) {
+    this.urls = serviceURLs(url); this.version = version; this.fetch = fetchImpl.bind(globalThis);
+    this.headers = { ...headers, 'A2A-Version': version };
+  }
+  method(v1, v03) { return this.version === '0.3' ? v03 : v1; }
+  async discover(signal) {
+    const response = await this.fetch(this.urls.card, { headers: this.headers, credentials: 'omit', signal });
+    if (!response.ok) throw new Error(`Não foi possível carregar o Agent Card (HTTP ${response.status}).`);
+    return response.json();
+  }
+  async rpc(method, params, { signal, timeoutMs = 15000 } = {}) {
+    const id = uid();
+    const timeout = AbortSignal.timeout(timeoutMs);
+    const response = await this.fetch(this.urls.rpc, {
+      method: 'POST', headers: { ...this.headers, 'Content-Type': 'application/json' }, credentials: 'omit',
+      body: JSON.stringify({ jsonrpc: '2.0', id, method, params }),
+      signal: signal ? AbortSignal.any([signal, timeout]) : timeout,
+    });
+    if (!response.ok) throw new Error(`O serviço retornou HTTP ${response.status}. Confira o endereço e a autenticação.`);
+    const data = await response.json();
+    return this.validate(data, id);
+  }
+  validate(data, id) {
+    if (data.jsonrpc !== '2.0' || data.id !== id) throw new Error('Resposta JSON-RPC inválida.');
+    if (data.error) throw new RpcError(data.error);
+    if (!Object.hasOwn(data, 'result')) throw new Error('Resposta sem resultado A2A.');
+    return data.result;
+  }
+  async sendMessage(text, { contextId, taskId, tenantId, agentId, backendId, data = {}, signal } = {}) {
+    const legacy = this.version === '0.3';
+    const message = {
+      messageId: uid(), role: legacy ? 'user' : 'ROLE_USER', contextId,
+      parts: [{ text }, { data }], metadata: {},
+    };
+    if (taskId) message.taskId = taskId;
+    for (const [key, value] of Object.entries({ tenant_id: tenantId, agent_id: agentId, backend_id: backendId })) {
+      if (value) message.metadata[key] = value;
+    }
+    if (legacy) {
+      message.kind = 'message'; message.parts[0].kind = 'text'; message.parts[1].kind = 'data';
+    }
+    const result = await this.rpc(this.method('SendMessage', 'message/send'), {
+      message, configuration: legacy ? { blocking: false } : { returnImmediately: true },
+    }, { signal });
+    return normalizeResult(result);
+  }
+  async getTask(id, signal) { return this.rpc(this.method('GetTask', 'tasks/get'), { id }, { signal }); }
+  async cancelTask(id) { return this.rpc(this.method('CancelTask', 'tasks/cancel'), { id }); }
+  async *subscribe(id, signal) {
+    const requestId = uid();
+    const response = await this.fetch(this.urls.rpc, {
+      method: 'POST', credentials: 'omit', signal,
+      headers: { ...this.headers, 'Content-Type': 'application/json', Accept: 'text/event-stream' },
+      body: JSON.stringify({ jsonrpc: '2.0', id: requestId, method: this.method('SubscribeToTask', 'tasks/resubscribe'), params: { id } }),
+    });
+    if (!response.ok) throw new Error(`Não foi possível acompanhar os eventos (HTTP ${response.status}).`);
+    if (!(response.headers.get('content-type') || '').includes('text/event-stream')) {
+      this.validate(await response.json(), requestId);
+      throw new Error('O serviço não retornou um fluxo SSE.');
+    }
+    for await (const frame of parseSSE(response.body)) {
+      yield normalizeResult(this.validate(JSON.parse(frame), requestId));
+    }
+  }
+}
+
+export class TaskView {
+  constructor(task = {}) { this.task = structuredClone(task); this.task.artifacts ||= []; }
+  apply(event) {
+    const value = event.value;
+    if (event.type === 'task') {
+      if (this.task.id && value.id !== this.task.id) throw new Error('Evento de outra tarefa.');
+      this.task = structuredClone(value); this.task.artifacts ||= [];
+    }
+    else if (event.type === 'status') {
+      if (this.task.id && value.taskId !== this.task.id) throw new Error('Evento de outra tarefa.');
+      this.task.id = value.taskId; this.task.contextId = value.contextId; this.task.status = value.status;
+    } else if (event.type === 'artifact') {
+      if (this.task.id && value.taskId !== this.task.id) throw new Error('Evento de outra tarefa.');
+      const index = this.task.artifacts.findIndex(a => a.artifactId === value.artifact.artifactId);
+      if (index < 0) this.task.artifacts.push(structuredClone(value.artifact));
+      else if (value.append) this.task.artifacts[index].parts.push(...structuredClone(value.artifact.parts || []));
+      else this.task.artifacts[index] = structuredClone(value.artifact);
+    }
+    return this.task;
+  }
+}
+
+export function displayResponse(value, depth = 0) {
+  if (!value || depth > 5) return null;
+  const task = value.task || value;
+  const parts = task.artifacts?.at(-1)?.parts || task.status?.message?.parts || value.message?.parts || value.parts || [];
+  const native = parts.find(p => p.data && typeof p.data === 'object' && ('text' in p.data || 'speak' in p.data || 'metadata' in p.data))?.data;
+  if (native) {
+    const remote = native.metadata?.a2a_result;
+    return (remote && displayResponse(remote, depth + 1)) || native;
+  }
+  const text = parts.filter(p => typeof p.text === 'string').map(p => p.text).join('');
+  if (text) return { text };
+  const structured = parts.filter(p => 'data' in p).map(p => p.data);
+  return structured.length ? { text: JSON.stringify(structured.length === 1 ? structured[0] : structured, null, 2) } : null;
+}

apps/agent_frontend_a2a/app.js

@@ -0,0 +1,225 @@
+import { A2AClient, TaskView, displayResponse, isStopped, needsInput, selectVersion, stateName, uid } from './a2a-client.js';
+
+const $ = id => document.getElementById(id);
+const value = id => $(id).value.trim();
+const labels = { submitted:'Tarefa recebida', working:'Agente trabalhando', completed:'Concluída', failed:'Falha no processamento', canceled:'Tarefa cancelada', rejected:'Solicitação recusada', 'input-required':'Aguardando sua resposta', 'auth-required':'Aguardando autorização' };
+const eventLabels = { 'flow.start':'Fluxo iniciado', 'session.upserted':'Sessão preparada', 'message.received':'Mensagem recebida', 'workflow.started':'Workflow iniciado', 'workflow.completed':'Workflow concluído', 'message.responded':'Resposta disponível', 'flow.end':'Fluxo finalizado', 'a2a.backend.status':'Status do backend atualizado' };
+const state = { client:null, card:null, contextId:uid(), view:new TaskView(), connected:false, busy:false, epoch:0, controller:null, turn:null, events:0, lastState:'', scope:null };
+const savedKey = 'framework-a2a-session-v1';
+const connectionFields = ['serviceUrl','target','version','skill','tenant','tracking','customerKey','contractKey','agentOverride','authHeader','authValue'];
+const messageGroups = new Map();
+
+function notice(text = '') { $('notice').textContent = text; $('notice').hidden = !text; }
+function setStatus(text, kind = '') { $('taskStatus').textContent = text; $('taskDot').className = `dot ${kind}`; }
+function setBusy(busy) { state.busy = busy; document.body.setAttribute('aria-busy', String(busy)); controls(); }
+function controls() {
+  const active = state.view.task.id && !isStopped(state.view.task.status?.state);
+  const allowSend = state.connected && !state.busy && !active;
+  $('message').disabled = !allowSend; $('sendButton').disabled = !allowSend;
+  $('cancelButton').disabled = !state.connected || !(active || needsInput(state.view.task.status?.state));
+  $('resumeButton').disabled = !state.connected || state.busy || !state.view.task.id;
+  $('newConversation').disabled = state.busy;
+  $('connectButton').disabled = state.busy;
+  for (const id of connectionFields) $(id).disabled = state.busy;
+}
+function log(text) {
+  const item = document.createElement('li');
+  const time = document.createElement('time'); time.textContent = new Date().toLocaleTimeString('pt-BR');
+  item.append(time, document.createTextNode(text)); $('activity').append(item);
+  while ($('activity').children.length > 80) $('activity').firstChild.remove();
+  $('eventCount').textContent = `${++state.events} eventos`;
+}
+function saveSession() {
+  // Only identifiers and connection settings: no credentials or message content.
+  try { sessionStorage.setItem(savedKey, JSON.stringify({
+    serviceUrl: state.client?.urls.base, version:state.client?.version, target:value('target'),
+    contextId:state.contextId, taskId:state.view.task.id, taskState:state.view.task.status?.state,
+    tenant:value('tenant'), skill:value('skill'), agentOverride:value('agentOverride'),
+  })); } catch { /* Storage can be unavailable in private browsing. */ }
+}
+function ids() {
+  $('taskId').textContent = state.view.task.id || '—'; $('taskShort').textContent = state.view.task.id ? `#${state.view.task.id.slice(0,8)}` : '';
+  $('contextId').textContent = state.contextId; $('contextShort').textContent = `Conversa ${state.contextId.slice(0,8)}`;
+}
+function stopWatcher() { ++state.epoch; state.controller?.abort(); state.controller = null; }
+function newConversation() {
+  stopWatcher(); state.contextId = uid(); state.view = new TaskView(); state.turn = null; state.lastState = ''; messageGroups.clear();
+  $('chat').replaceChildren(); $('activity').replaceChildren(); state.events = 0; $('eventCount').textContent='0 eventos'; notice();
+  setBusy(false); ids(); setStatus(state.connected ? 'Pronto para conversar' : 'Pronto para conectar'); saveSession(); $('message').focus();
+}
+function userMessage(text) {
+  const div = document.createElement('div'); div.className='message user'; div.textContent=text;
+  $('welcome')?.remove(); $('chat').append(div); scrollChat();
+}
+function agentMessages(response) {
+  const metadata = response.metadata || {};
+  const candidates = response.supervisor_results || metadata.supervisor_results || response.agent_responses || metadata.agent_responses;
+  if (Array.isArray(candidates)) {
+    const parts = candidates.map(p => ({ agent:p.agent || p.agent_id || p.name || 'Agente', text:p.answer ?? p.text ?? p.message ?? p.response ?? p.content, primary:p.primary }));
+    const valid = parts.filter(p => p.text != null && String(p.text).trim());
+    if (valid.length) return valid.sort((a,b)=>Number(Boolean(a.primary))-Number(Boolean(b.primary)));
+  }
+  const text = response.text ?? response.speak ?? response.content ?? '';
+  return [{ agent:metadata.active_agent || metadata.agent_id || state.card?.name || 'Agente', text }];
+}
+function scrollChat() { $('chat').scrollTop = $('chat').scrollHeight; }
+function renderResponse(response) {
+  if (!response) return;
+  $('welcome')?.remove(); state.turn ||= uid();
+  let group = messageGroups.get(state.turn);
+  if (!group) { group=document.createElement('section'); group.className='message agent'; group.setAttribute('aria-label','Resposta do agente'); messageGroups.set(state.turn,group); $('chat').append(group); }
+  const signature=JSON.stringify(response); if (group.dataset.signature === signature) return;
+  group.dataset.signature=signature; group.replaceChildren();
+  for (const part of agentMessages(response)) {
+    const section=document.createElement('section'); section.className='agent-section';
+    const header=document.createElement('div'); header.className='agent-header';
+    const avatar=document.createElement('span'); avatar.className='agent-avatar'; avatar.textContent='O'; avatar.setAttribute('aria-hidden','true');
+    const name=document.createElement('span'); name.textContent=String(part.agent).replaceAll('_',' ');
+    const body=document.createElement('div'); body.className='agent-body'; body.textContent=typeof part.text==='string' ? part.text : JSON.stringify(part.text,null,2);
+    header.append(avatar,name); section.append(header,body); group.append(section);
+  }
+  scrollChat();
+}
+function updateTask(event) {
+  const task=state.view.apply(event);
+  if (task.contextId) state.contextId=task.contextId;
+  const current=stateName(task.status?.state);
+  if (current && current!==state.lastState) { log(labels[current] || current); state.lastState=current; }
+  if (event.type==='status' && event.value.metadata?.framework_event) log(eventLabels[event.value.metadata.framework_event] || event.value.metadata.framework_event);
+  setStatus(labels[current] || 'Acompanhando tarefa', ['failed','rejected'].includes(current) ? 'error' : isStopped(current) ? '' : 'busy');
+  if (event.type==='artifact' || isStopped(current)) renderResponse(displayResponse(task));
+  ids(); saveSession(); controls();
+}
+function sleep(ms,signal) {
+  return new Promise((resolve,reject)=>{
+    if (signal.aborted) { reject(new DOMException('Interrompido','AbortError')); return; }
+    const abort=()=>{clearTimeout(timer);reject(new DOMException('Interrompido','AbortError'));};
+    const timer=setTimeout(()=>{signal.removeEventListener('abort',abort);resolve();},ms);
+    signal.addEventListener('abort',abort,{once:true});
+  });
+}
+async function watchTask() {
+  stopWatcher(); const epoch=state.epoch; const controller=new AbortController(); state.controller=controller;
+  const signal=controller.signal; const id=state.view.task.id; setBusy(true);
+  try {
+    if (value('tracking')==='stream' && state.card?.capabilities?.streaming && !isStopped(state.view.task.status?.state)) {
+      try {
+        log('Conectando ao fluxo da tarefa');
+        for await (const event of state.client.subscribe(id,signal)) {
+          if (epoch!==state.epoch) return; updateTask(event);
+          if (isStopped(state.view.task.status?.state)) break;
+        }
+      } catch (error) {
+        if (signal.aborted) return;
+        log('Fluxo interrompido; recuperando por consulta');
+      }
+    }
+    // Always obtain the authoritative final snapshot; also recovers a fast Task
+    // that completed before SubscribeToTask or a dropped/truncated SSE stream.
+    let failures=0;
+    while (true) {
+      try {
+        const task=await state.client.getTask(id,signal);
+        if (epoch!==state.epoch) return;
+        updateTask({type:'task',value:task}); failures=0;
+        if (isStopped(task.status?.state)) break;
+      } catch (error) {
+        if (signal.aborted) return;
+        if (++failures>=3 || error.code) throw error;
+        log('Conexão indisponível; tentando consultar novamente');
+      }
+      await sleep(failures ? failures*1000 : 900,signal);
+    }
+  } catch (error) {
+    if (!signal.aborted && epoch===state.epoch) {
+      notice(`Acompanhamento pausado: ${error.message} Use Retomar para consultar a mesma tarefa. Nenhuma mensagem foi reenviada.`);
+      setStatus('Acompanhamento pausado','error'); log('Acompanhamento pausado');
+    }
+  } finally {
+    if (epoch===state.epoch) { setBusy(false); $('message').focus(); }
+  }
+}
+function authHeaders() {
+  const secret=$('authValue').value;
+  if (!secret) return {};
+  const name=value('authHeader');
+  if (!/^[!#$%&'*+.^_`|~0-9A-Za-z-]+$/.test(name) || /[\r\n]/.test(secret)) throw new Error('Cabeçalho de autenticação inválido.');
+  if (['content-type','accept','a2a-version','host','origin','cookie'].includes(name.toLowerCase())) throw new Error('Use Authorization ou o cabeçalho de autenticação configurado no serviço.');
+  return {[name]:secret};
+}
+$('connectionForm').addEventListener('submit',async event=>{
+  event.preventDefault(); notice(); setBusy(true); setStatus('Conectando…','busy');
+  try {
+    const candidate=new A2AClient({url:value('serviceUrl'),version:value('version')==='0.3'?'0.3':'1.0',headers:authHeaders()});
+    const card=await candidate.discover(AbortSignal.timeout(12000)); candidate.version=selectVersion(card,value('version')); candidate.headers['A2A-Version']=candidate.version;
+    const scope=JSON.stringify([candidate.urls.base,candidate.version,value('target'),value('tenant'),value('skill'),value('agentOverride')]);
+    if (state.scope && scope!==state.scope) newConversation();
+    state.scope=scope; state.client=candidate; state.card=card; state.connected=true;
+    const previous=value('skill'); $('skill').replaceChildren(new Option('Seleção automática',''));
+    for (const skill of card.skills || []) $('skill').append(new Option(skill.name || skill.id,skill.id));
+    $('skill').value=[...$('skill').options].some(o=>o.value===previous)?previous:'';
+    $('serviceName').textContent=card.name || 'Serviço A2A'; $('connectionDetail').textContent=`${(card.skills || []).length === 1 ? '1 opção disponível' : `${(card.skills || []).length} opções disponíveis`} · ${card.capabilities?.streaming ? 'Atualizações em tempo real' : 'Consulta de tarefas'}`;
+    $('connectionDot').className='dot'; $('versionBadge').textContent=`A2A ${candidate.version}`;
+    setStatus(state.view.task.id?'Conectado · tarefa disponível para consulta':'Pronto para conversar');
+    log('Conectado ao serviço'); ids(); saveSession();
+  } catch(error) {
+    state.connected=false; $('connectionDot').className='dot error'; $('connectionDetail').textContent='Não foi possível conectar.';
+    setStatus('Falha na conexão','error'); notice(`${error.message} Se o navegador bloquear a conexão, confira CORS_ORIGINS no serviço e use HTTP/HTTPS para servir esta página.`);
+  } finally { setBusy(false); }
+});
+$('messageForm').addEventListener('submit',async event=>{
+  event.preventDefault(); const text=value('message'); if(!text || !state.connected || $('sendButton').disabled) return;
+  notice(); const previousTask=state.view.task; const resumeId=needsInput(previousTask.status?.state)?previousTask.id:undefined;
+  state.turn=uid(); userMessage(text); $('message').value=''; $('message').style.height='auto'; setBusy(true); setStatus('Enviando mensagem…','busy');
+  const messageId=uid();
+  const business_context={customer_key:value('customerKey')||null,contract_key:value('contractKey')||null,interaction_key:messageId,session_key:state.contextId};
+  const data={message_id:messageId,business_context,context:{business_context}};
+  if(value('customerKey')) data.customer_key=value('customerKey');
+  if(value('contractKey')) data.contract_key=value('contractKey');
+  try {
+    const gateway=value('target')==='gateway';
+    const result=await state.client.sendMessage(text,{contextId:state.contextId,taskId:resumeId,tenantId:value('tenant')||'default',
+      agentId:gateway?value('agentOverride'):value('skill'),backendId:gateway?value('skill'):undefined,data});
+    if(result.type==='message') {
+      state.view=new TaskView(); if(result.value.contextId) state.contextId=result.value.contextId;
+      renderResponse(displayResponse(result.value)); setStatus('Resposta recebida'); ids(); saveSession(); setBusy(false);
+    } else if(result.type==='task') {
+      state.view=new TaskView(); state.lastState=''; updateTask(result); log(`Tarefa ${state.view.task.id.slice(0,8)} recebida`); await watchTask();
+    } else throw new Error('Resposta de envio inesperada.');
+  } catch(error) {
+    notice(`Não foi possível confirmar o envio: ${error.message} Ele pode ter sido aceito pelo serviço; não houve reenvio automático.`);
+    setStatus('Envio não confirmado','error'); setBusy(false);
+  }
+});
+$('resumeButton').addEventListener('click',()=>{notice();state.turn ||= uid();watchTask();});
+$('cancelButton').addEventListener('click',async()=>{
+  const id=state.view.task.id; if(!id || !state.connected) return;
+  stopWatcher(); setBusy(true); setStatus('Solicitando cancelamento…','busy');
+  try { updateTask({type:'task',value:await state.client.cancelTask(id)});log('Cancelamento confirmado pelo serviço');notice(); }
+  catch(error) {
+    try {updateTask({type:'task',value:await state.client.getTask(id)});} catch { /* Keep the last known Task. */ }
+    notice(`O serviço não confirmou o cancelamento: ${error.message} Consulte a tarefa com Retomar.`);
+  } finally {setBusy(false);}
+});
+$('newConversation').addEventListener('click',newConversation);
+$('message').addEventListener('keydown',event=>{if(event.key==='Enter'&&!event.shiftKey&&!event.isComposing){event.preventDefault();$('messageForm').requestSubmit();}});
+$('message').addEventListener('input',()=>{ $('message').style.height='auto';$('message').style.height=`${Math.min($('message').scrollHeight,150)}px`; });
+for(const button of document.querySelectorAll('[data-prompt]')) button.addEventListener('click',()=>{if(!$('message').disabled){$('message').value=button.dataset.prompt;$('message').focus();}else notice('Conecte ao serviço para iniciar a conversa.');});
+$('target').addEventListener('change',()=>{ $('skillLabel').firstChild.textContent=value('target')==='gateway'?'Backend no Gateway':'Agente'; });
+for(const id of ['serviceUrl','version','target','tenant','skill','agentOverride','authHeader','authValue']) $(id).addEventListener('change',()=>{
+  if(state.connected){state.connected=false;$('connectionDot').className='dot muted';setStatus('Configuração alterada · conecte novamente');controls();}
+});
+try {
+  const saved=JSON.parse(sessionStorage.getItem(savedKey)||'null');
+  if(saved?.serviceUrl){
+    $('serviceUrl').value=saved.serviceUrl;$('version').value=saved.version||'auto';$('target').value=saved.target||'backend';$('tenant').value=saved.tenant||'default';$('agentOverride').value=saved.agentOverride||'';
+    if(saved.skill)$('skill').append(new Option(saved.skill,saved.skill));$('skill').value=saved.skill||'';
+    state.contextId=saved.contextId||uid();state.view=new TaskView(saved.taskId?{id:saved.taskId,contextId:state.contextId,status:{state:saved.taskState}}:{});
+    state.scope=JSON.stringify([saved.serviceUrl,saved.version,saved.target||'backend',saved.tenant||'default',saved.skill||'',saved.agentOverride||'']);
+    if(saved.taskId)notice('Uma tarefa desta aba está disponível. Conecte ao mesmo serviço e clique em Retomar para consultar o resultado.');
+  }
+} catch { /* An unavailable or invalid saved session starts a fresh conversation. */ }
+const requestedService=new URLSearchParams(location.search).get('service');
+if(requestedService)$('serviceUrl').value=requestedService;
+else if(!state.scope && location.port==='8000')$('serviceUrl').value=location.origin;
+$('target').dispatchEvent(new Event('change'));ids();controls();

apps/agent_frontend_a2a/index.html

@@ -0,0 +1,52 @@
+<!doctype html>
+<html lang="pt-BR">
+<head>
+  <meta charset="utf-8"><meta name="viewport" content="width=device-width,initial-scale=1">
+  <title>Oracle Agent Workspace · A2A</title><link rel="stylesheet" href="styles.css">
+</head>
+<body>
+  <div class="workspace">
+    <aside class="sidebar">
+      <a class="brand" href="./"><img class="oracle-logo" src="oracle-logo.svg" alt="Oracle"><span class="brand-copy">Agent Workspace<small>FRAMEWORK OCI · A2A</small></span></a>
+      <div class="sidebar-heading"><span>CONEXÃO</span><span class="badge" id="versionBadge">A2A</span></div>
+      <form id="connectionForm" class="connection-form">
+        <label>Endereço do serviço<input id="serviceUrl" type="url" value="http://localhost:8000" required spellcheck="false"></label>
+        <div class="two-columns">
+          <label>Destino<select id="target" aria-label="Destino"><option value="backend">Agente</option><option value="gateway">Gateway</option></select></label>
+          <label>Versão<select id="version" aria-label="Versão"><option value="auto">Automática</option><option value="1.0">A2A 1.0</option><option value="0.3">A2A 0.3</option></select></label>
+        </div>
+        <button id="connectButton" class="primary" type="submit">Conectar ao serviço <span aria-hidden="true">↗</span></button>
+      </form>
+      <div class="service-card"><span id="connectionDot" class="dot"></span><div><strong id="serviceName">Nenhum serviço conectado</strong><p id="connectionDetail">Conecte para carregar os agentes disponíveis.</p></div></div>
+      <div class="sidebar-heading"><span>CONVERSA</span></div>
+      <label id="skillLabel">Agente<select id="skill" aria-label="Agente ou backend"><option value="">Seleção automática</option></select></label>
+      <label>Tenant<input id="tenant" value="default" spellcheck="false"></label>
+      <label>Acompanhamento<select id="tracking" aria-label="Acompanhamento"><option value="stream">Tempo real</option><option value="poll">Consulta periódica</option></select></label>
+      <details class="advanced"><summary>Contexto e autenticação</summary>
+        <label>Identificador do cliente<input id="customerKey" placeholder="customer_key" autocomplete="off"></label>
+        <label>Contrato ou pedido<input id="contractKey" placeholder="contract_key" autocomplete="off"></label>
+        <label>Perfil do agente no backend<input id="agentOverride" placeholder="Opcional; usado pelo Gateway" autocomplete="off"></label>
+        <label>Cabeçalho de autenticação<input id="authHeader" value="Authorization" autocomplete="off" spellcheck="false"></label>
+        <label>Valor do cabeçalho<input id="authValue" type="password" placeholder="Bearer … ou credencial configurada" autocomplete="off"></label>
+        <p class="hint">A credencial fica apenas na memória desta página.</p>
+      </details>
+      <div class="sidebar-footer"><span class="dot muted"></span> Protocolo A2A · JSON-RPC + SSE</div>
+    </aside>
+    <main class="main">
+      <header class="topbar"><div><span class="eyebrow">SEU ESPAÇO DE TRABALHO</span><h1>Converse com seus agentes</h1></div><button id="newConversation" class="secondary" type="button">＋ Nova conversa</button></header>
+      <section class="task-strip" aria-label="Acompanhamento da tarefa">
+        <div class="status-pill"><span id="taskDot" class="dot muted"></span><span id="taskStatus" role="status" aria-live="polite">Pronto para conectar</span></div>
+        <span id="taskShort" class="task-id"></span>
+        <div class="task-actions"><button id="resumeButton" class="text-button" type="button" disabled>Retomar</button><button id="cancelButton" class="text-button danger" type="button" disabled>Cancelar tarefa</button></div>
+      </section>
+      <div id="notice" class="notice" role="alert" hidden></div>
+      <section id="chat" class="chat" role="log" aria-label="Conversa com o agente" aria-live="polite">
+        <div id="welcome" class="welcome"><span class="eyebrow">CONVERSAS QUE VIRAM AÇÕES</span><h2>O próximo passo<br>começa com uma mensagem.</h2><p>Conecte um agente ou o Gateway do framework.<br>Acompanhe cada tarefa até a resposta.</p><div class="suggestions"><button type="button" data-prompt="O que você pode fazer por mim?">Conhecer o agente <span>↗</span></button><button type="button" data-prompt="Quero consultar meu pedido.">Consultar um pedido <span>↗</span></button></div></div>
+      </section>
+      <div class="composer-area"><form id="messageForm" class="composer"><label class="sr-only" for="message">Mensagem para o agente</label><textarea id="message" rows="1" placeholder="Escreva uma mensagem para começar…" disabled></textarea><button id="sendButton" type="submit" class="send-button" aria-label="Enviar mensagem" disabled>↑</button></form><div class="composer-footer"><span>Enter envia · Shift + Enter quebra a linha</span><span id="contextShort">Conversa nova</span></div></div>
+      <details class="activity-panel"><summary>Atividade da tarefa <span id="eventCount">0 eventos</span></summary><ol id="activity"></ol><dl><dt>Contexto</dt><dd id="contextId">—</dd><dt>Tarefa</dt><dd id="taskId">—</dd></dl></details>
+    </main>
+  </div>
+  <script type="module" src="app.js"></script>
+</body>
+</html>

apps/agent_frontend_a2a/oracle-logo.svg

@@ -0,0 +1,2 @@
+<?xml version="1.0" encoding="utf-8"?><!-- Uploaded to: SVG Repo, www.svgrepo.com, Generator: SVG Repo Mixer Tools -->
+<svg width="800px" height="800px" viewBox="0 0 16 16" xmlns="http://www.w3.org/2000/svg" fill="none"><path fill="#EA1B22" fill-rule="evenodd" d="M.1 8c0 2.761 2.237 5 4.997 5h5.806A4.999 4.999 0 0015.9 8c0-2.761-2.237-5-4.997-5H5.097A4.999 4.999 0 00.1 8zm13.911 0a3.235 3.235 0 01-3.234 3.237h-5.55A3.235 3.235 0 011.991 8a3.235 3.235 0 013.234-3.236h5.551A3.235 3.235 0 0114.011 8z" clip-rule="evenodd"/></svg>

apps/agent_frontend_a2a/package.json

@@ -0,0 +1 @@
+{"name":"agent-framework-a2a-frontend","private":true,"type":"module","scripts":{"test":"node --test tests/*.test.js"}}

apps/agent_frontend_a2a/tests/client.test.js

@@ -0,0 +1,86 @@
+import test from 'node:test';
+import assert from 'node:assert/strict';
+import { A2AClient, TaskView, displayResponse, parseSSE, serviceURLs, selectVersion, normalizeResult, isStopped, needsInput, RpcError } from '../a2a-client.js';
+
+function bodyFor(text, size = 1) {
+  const bytes=new TextEncoder().encode(text);
+  return new ReadableStream({start(controller){for(let i=0;i<bytes.length;i+=size)controller.enqueue(bytes.slice(i,i+size));controller.close();}});
+}
+
+test('SSE handles fragmented UTF-8, CRLF, comments and multi-line data',async()=>{
+  const text=': ping\r\nid: 1\r\nevent: message\r\ndata: {"text":\r\ndata: "Olá 👋"}\r\n\r\ndata: second\n\n';
+  const frames=[];for await(const frame of parseSSE(bodyFor(text)))frames.push(frame);
+  assert.deepEqual(frames,['{"text":\n"Olá 👋"}','second']);
+});
+test('SSE discards an unfinished frame for recovery by GetTask',async()=>{
+  const frames=[];for await(const frame of parseSSE(bodyFor('data: complete\n\ndata: truncated')))frames.push(frame);
+  assert.deepEqual(frames,['complete']);
+});
+test('service URLs accept a base or /a2a and reject embedded credentials',()=>{
+  assert.equal(serviceURLs('http://localhost:8000/a2a/').rpc,'http://localhost:8000/a2a');
+  assert.equal(serviceURLs('https://host/prefix').card,'https://host/prefix/.well-known/agent-card.json');
+  for(const url of ['javascript:alert(1)','https://user:secret@host','https://host?token=secret'])assert.throws(()=>serviceURLs(url));
+});
+test('version selection uses JSONRPC interfaces and allows explicit legacy compatibility',()=>{
+  assert.equal(selectVersion({supportedInterfaces:[{protocolBinding:'JSONRPC',protocolVersion:'1.0'}]}),'1.0');
+  assert.equal(selectVersion({protocolVersion:'0.3.0'}),'0.3');
+  assert.equal(selectVersion({},'0.3'),'0.3');
+  assert.throws(()=>selectVersion({supportedInterfaces:[{protocolBinding:'GRPC',protocolVersion:'1.0'}]}));
+});
+test('both versions normalize task, message and streaming events',()=>{
+  assert.equal(normalizeResult({statusUpdate:{taskId:'t'}}).type,'status');
+  assert.equal(normalizeResult({kind:'artifact-update',taskId:'t'}).type,'artifact');
+  assert.equal(normalizeResult({message:{messageId:'m'}}).type,'message');
+  assert.throws(()=>normalizeResult({unexpected:true}));
+});
+test('TaskView updates status, replaces snapshots and appends artifact chunks',()=>{
+  const view=new TaskView({id:'t',artifacts:[]});
+  view.apply({type:'artifact',value:{taskId:'t',artifact:{artifactId:'a',parts:[{text:'Hel'}]}}});
+  view.apply({type:'artifact',value:{taskId:'t',append:true,artifact:{artifactId:'a',parts:[{text:'lo'}]}}});
+  assert.equal(displayResponse(view.task).text,'Hello');
+  view.apply({type:'status',value:{taskId:'t',contextId:'c',status:{state:'TASK_STATE_COMPLETED'}}});
+  assert.ok(isStopped(view.task.status.state));assert.ok(needsInput('input-required'));
+  assert.throws(()=>view.apply({type:'status',value:{taskId:'other',status:{state:'working'}}}));
+});
+test('framework text/data duplicates render once and native agent metadata survives',()=>{
+  const native={text:'answer',metadata:{agent_responses:[{agent:'billing',text:'answer'}]}};
+  const task={artifacts:[{parts:[{text:'answer'},{data:native}]}],status:{message:{parts:[{text:'answer'}]}}};
+  assert.deepEqual(displayResponse(task),native);
+  const wrapped={parts:[{data:{text:'answer',metadata:{a2a_result:{task}}}}]};
+  assert.deepEqual(displayResponse(wrapped),native);
+});
+for(const legacy of [false,true]){
+  test(`send/get/cancel use correct envelopes and credentials (${legacy?'0.3':'1.0'})`,async()=>{
+    const requests=[];
+    const fetchImpl=async function(url,options){
+      assert.equal(this,globalThis);
+      const body=JSON.parse(options.body);requests.push({url,options,body});
+      return Response.json({jsonrpc:'2.0',id:body.id,result:body.method.includes('send')||body.method==='SendMessage' ? (legacy?{kind:'task',id:'t',status:{state:'working'}}:{task:{id:'t',status:{state:'TASK_STATE_WORKING'}}}):{id:'t'}});
+    };
+    const client=new A2AClient({url:'https://service',version:legacy?'0.3':'1.0',headers:{Authorization:'Bearer test'},fetchImpl});
+    await client.sendMessage('hello',{contextId:'c',agentId:'a',backendId:'billing',tenantId:'tenant',data:{business_context:{customer_key:'customer'}}});
+    await client.getTask('t');await client.cancelTask('t');
+    assert.deepEqual(requests.map(r=>r.body.method),legacy?['message/send','tasks/get','tasks/cancel']:['SendMessage','GetTask','CancelTask']);
+    const params=requests[0].body.params;
+    assert.equal(params.message.role,legacy?'user':'ROLE_USER');assert.equal(params.message.metadata.backend_id,'billing');
+    assert.equal(params.message.metadata.agent_id,'a');assert.equal(params.message.contextId,'c');
+    assert.equal(legacy?params.configuration.blocking:params.configuration.returnImmediately,!legacy);
+    assert.equal(requests[0].options.headers.Authorization,'Bearer test');
+    assert.equal(requests[0].options.credentials,'omit');
+  });
+}
+test('JSON-RPC errors and mismatched response IDs never become successful responses',async()=>{
+  const bad=new A2AClient({url:'https://service',fetchImpl:async()=>Response.json({jsonrpc:'2.0',id:'wrong',result:{}})});
+  await assert.rejects(bad.getTask('t'),/JSON-RPC/);
+  const failure=new A2AClient({url:'https://service',fetchImpl:async(url,o)=>Response.json({jsonrpc:'2.0',id:JSON.parse(o.body).id,error:{code:-32001,message:'Task missing'}})});
+  await assert.rejects(failure.getTask('t'),RpcError);
+});
+test('POST SSE supports auth headers, validates envelopes and streams canonical events',async()=>{
+  const client=new A2AClient({url:'https://service',headers:{'x-api-key':'test'},fetchImpl:async(url,o)=>{
+    assert.equal(o.method,'POST');assert.equal(o.headers['x-api-key'],'test');
+    const body=JSON.parse(o.body);assert.equal(body.method,'SubscribeToTask');
+    return new Response(bodyFor(`data: ${JSON.stringify({jsonrpc:'2.0',id:body.id,result:{statusUpdate:{taskId:'t',status:{state:'TASK_STATE_COMPLETED'}}}})}\n\n`),{headers:{'content-type':'text/event-stream'}});
+  }});
+  const events=[];for await(const event of client.subscribe('t'))events.push(event);
+  assert.equal(events[0].type,'status');assert.equal(events[0].value.taskId,'t');
+});

apps/agent_frontend_a2a/tests/integration.test.js

@@ -0,0 +1,32 @@
+// Run against examples/a2a_frontend_demo.py with A2A_DEMO_URL=http://127.0.0.1:8765.
+import test from 'node:test';
+import assert from 'node:assert/strict';
+import { A2AClient, TaskView, displayResponse, isStopped, stateName } from '../a2a-client.js';
+const url=process.env.A2A_DEMO_URL;
+const delay=ms=>new Promise(resolve=>setTimeout(resolve,ms));
+async function complete(client,task){
+  const deadline=Date.now()+6000;
+  while(!isStopped(task.status?.state)&&Date.now()<deadline){await delay(50);task=await client.getTask(task.id);}
+  assert.ok(isStopped(task.status?.state));return task;
+}
+for(const version of ['1.0','0.3']){
+  test(`SDK integration: submit, subscribe, poll and resume (${version})`,{skip:!url},async()=>{
+    const client=new A2AClient({url,version});const card=await client.discover();assert.ok(card.capabilities.streaming);
+    const first=await client.sendMessage('hello',{contextId:crypto.randomUUID()});assert.equal(first.type,'task');assert.ok(!isStopped(first.value.status.state));
+    const view=new TaskView(first.value);let count=0;
+    for await(const event of client.subscribe(first.value.id)){view.apply(event);count++;}
+    assert.ok(count>0);const final=await client.getTask(first.value.id);
+    assert.equal(stateName(final.status.state),'completed');assert.match(displayResponse(final).text,/hello/);
+    const prompt=await client.sendMessage('/confirmar',{contextId:crypto.randomUUID()});
+    const waiting=await complete(client,prompt.value);assert.equal(stateName(waiting.status.state),'input-required');
+    const resumed=await client.sendMessage('sim',{contextId:waiting.contextId,taskId:waiting.id});
+    const finished=await complete(client,resumed.value);assert.equal(finished.id,waiting.id);assert.equal(stateName(finished.status.state),'completed');
+  });
+}
+test('SDK integration: cancellation and failed tasks',{skip:!url},async()=>{
+  const client=new A2AClient({url});
+  const slow=await client.sendMessage('/lento',{contextId:crypto.randomUUID()});
+  await delay(50);const canceled=await client.cancelTask(slow.value.id);assert.equal(stateName(canceled.status.state),'canceled');
+  const failed=await client.sendMessage('/erro',{contextId:crypto.randomUUID()});
+  assert.equal(stateName((await complete(client,failed.value)).status.state),'failed');
+});

apps/agent_gateway/Dockerfile.a2a

@@ -0,0 +1,7 @@
+FROM python:3.12-slim
+WORKDIR /app
+COPY libs/agent_framework /agent_framework
+COPY apps/agent_gateway /app
+RUN pip install --no-cache-dir -e '/agent_framework[a2a]' -r requirements.txt
+ENV ENABLE_A2A=true
+CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8010"]

apps/agent_gateway/README.md

@@ -71,3 +71,13 @@ O Gateway emite eventos de observabilidade:
 - `IC.GLOBAL_GATEWAY_COMPLETED`
 - `NOC.005` em falhas
 - `NOC.006` em conclusão HTTP
+
+## A2A opcional
+
+A fachada A2A 1.0, a compatibilidade 0.3 e o catálogo misto de backends são
+explicados em [Integração A2A](../../docs/A2A_IMPLEMENTACAO.md).
+A ativação é opcional (`ENABLE_A2A=false` por padrão).
+
+A fachada também oferece [Tasks A2A assíncronas e streaming](../../docs/A2A_ASSINCRONO.md),
+com envio imediato, consulta, reconexão e cancelamento. O template de backend
+pode publicar o mesmo protocolo diretamente.

apps/agent_gateway/app/main.py

@@ -16,7 +16,7 @@ from agent_framework.global_supervisor import (
     GlobalSupervisorRouter,
     InMemoryGlobalSessionStore,
 )
-from agent_framework.llm.providers import create_llm
+from app.supervisor_llm import build_supervisor_llm
 from agent_framework.observability.observer import AgentObserver
 from agent_framework.security import install_authentication
 
@@ -38,7 +38,7 @@ app.add_middleware(
 registry = BackendRegistry.from_yaml(settings.BACKENDS_CONFIG_PATH)
 analytics = create_analytics_publisher(settings)
 observer = AgentObserver(analytics=analytics)
-llm = create_llm(settings)
+llm = build_supervisor_llm(settings) if settings.GLOBAL_ROUTING_MODE in {"supervisor", "hybrid"} else None
 session_store = InMemoryGlobalSessionStore(ttl_seconds=settings.GLOBAL_SESSION_TTL_SECONDS)
 router = GlobalSupervisorRouter(
     registry=registry,
@@ -219,6 +219,10 @@ import asyncio
 @app.get("/gateway/events/{session_id:path}")
 async def gateway_events(session_id: str):
 
+    existing = await session_store.get(session_id)
+    if existing and existing.active_backend and registry.get(existing.active_backend).protocol == "a2a":
+        raise HTTPException(status_code=409, detail="A2A backend: use message/send; native event proxy is unavailable")
+
     async def stream():
         yield (
             "event: connected\n"
@@ -250,6 +254,9 @@ async def gateway_events(session_id: str):
             return
 
         backend = registry.get(backend_id)
+        if backend.protocol == "a2a":
+            yield 'event: error\ndata: {"message":"Native event proxy is unavailable for A2A backends"}\n\n'
+            return
 
         backend_base_url = (
                 getattr(backend, "base_url", None)
@@ -326,3 +333,16 @@ async def gateway_events(session_id: str):
             "X-Accel-Buffering": "no",
         },
     )
+
+# Opt-in: existing deployments do not import or require the A2A SDK.
+if settings.ENABLE_A2A:
+    from agent_framework.a2a.server import install_a2a
+
+    async def _a2a_gateway_handler(data: dict):
+        return await gateway_message(GatewayRequest(**data))
+
+    install_a2a(
+        app, _a2a_gateway_handler, registry, settings.A2A_PUBLIC_URL,
+        settings.A2A_SECURITY_SCHEMES or None, settings.A2A_SECURITY or None,
+        task_mode=settings.A2A_TASK_MODE, timeout_seconds=settings.A2A_TASK_TIMEOUT_SECONDS,
+    )

apps/agent_gateway/app/settings.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 from functools import lru_cache
 from typing import Literal
 
+from pydantic import Field
 from pydantic_settings import BaseSettings, SettingsConfigDict
 
 
@@ -16,6 +17,13 @@ class GatewaySettings(BaseSettings):
     API_PORT: int = 8010
     CORS_ORIGINS: str = "http://localhost:5173"
 
+    ENABLE_A2A: bool = False
+    A2A_TASK_MODE: bool = True
+    A2A_TASK_TIMEOUT_SECONDS: float = Field(default=120.0, gt=0)
+    A2A_PUBLIC_URL: str = "http://localhost:8010"
+    A2A_SECURITY_SCHEMES: dict = Field(default_factory=dict)
+    A2A_SECURITY: list[dict[str, list[str]]] = Field(default_factory=list)
+
     BACKENDS_CONFIG_PATH: str = "./config/backends.yaml"
     GLOBAL_ROUTING_MODE: Literal["router", "supervisor", "hybrid"] = "hybrid"
     GLOBAL_KEEP_ACTIVE_BACKEND: bool = True
@@ -38,6 +46,13 @@ class GatewaySettings(BaseSettings):
     LANGFUSE_HOST: str = "https://cloud.langfuse.com"
     MODEL_PRICES_JSON: str | None = None
     USD_BRL_RATE: str = "5.0"
+    OCI_COMPARTMENT_ID: str | None = None
+    OCI_GENAI_PROJECT_OCID: str | None = None
+
+    OCI_AUTH_MODE: str = "config_file"
+    OCI_CONFIG_FILE: str = "~/.oci/config"
+    OCI_CONFIG_PROFILE: str = "DEFAULT"
+    OCI_REGION: str | None = None
 
     # Analytics/Observer do próprio gateway.
     ENABLE_ANALYTICS: bool = False

apps/agent_gateway/app/supervisor_llm.py

@@ -0,0 +1,102 @@
+"""Gateway-only bootstrap for the Global Supervisor LLM.
+
+Resolve the supervisor entry in llm_profiles.yaml *before* create_llm().
+This matters because create_llm() selects OCISDKProvider immediately when
+LLM_PROVIDER=oci_sdk; OCISDKProvider does not dynamically switch providers.
+"""
+from __future__ import annotations
+
+import logging
+import os
+from pathlib import Path
+
+import yaml
+
+from agent_framework.llm.providers import create_llm
+
+logger = logging.getLogger("agent_gateway.supervisor_llm")
+
+
+def _profile_file(settings) -> Path:
+    explicit = (
+        os.getenv("GLOBAL_SUPERVISOR_LLM_PROFILES_PATH")
+        or getattr(settings, "GLOBAL_SUPERVISOR_LLM_PROFILES_PATH", None)
+        or os.getenv("LLM_PROFILES_PATH")
+        or getattr(settings, "LLM_PROFILES_PATH", None)
+    )
+    if explicit:
+        path = Path(str(explicit)).expanduser().resolve()
+        if not path.is_file():
+            raise RuntimeError(f"Supervisor LLM profile YAML not found: {path}")
+        return path
+
+    candidates = (
+        "config/llm_profiles.yaml",
+        "llm_profiles.yaml",
+        "apps/agent_gateway/config/llm_profiles.yaml",
+        "apps/agent_template_backend/config/llm_profiles.yaml",
+        "apps/agent_template_backend/llm_profiles.yaml",
+    )
+    for candidate in candidates:
+        path = Path(candidate)
+        if path.is_file():
+            return path.resolve()
+    raise RuntimeError(
+        "No llm_profiles.yaml found for Gateway supervisor. Set "
+        "GLOBAL_SUPERVISOR_LLM_PROFILES_PATH to the EXISTING YAML file path."
+    )
+
+
+def build_supervisor_llm(settings):
+    """Create LLM from the supervisor profile without mutating global settings."""
+    path = _profile_file(settings)
+    data = yaml.safe_load(path.read_text(encoding="utf-8")) or {}
+    profile = (data.get("profiles") or {}).get("supervisor")
+    if not isinstance(profile, dict):
+        raise RuntimeError(f"Missing profiles.supervisor in {path}")
+
+    effective_provider = str(profile.get("provider") or settings.LLM_PROVIDER).strip()
+    allowed = {"oci_openai", "openai_compatible", "oci_sdk", "mock"}
+    if effective_provider not in allowed:
+        raise RuntimeError(f"Unsupported supervisor provider: {effective_provider}")
+
+    configured_url = (
+        profile.get("base_url")
+        or profile.get("service_endpoint")
+        or getattr(settings, "OCI_GENAI_BASE_URL", None)
+    )
+    url = str(configured_url or "").strip().rstrip("/")
+    if effective_provider == "oci_openai":
+        # OCI native URL is commonly used by the process-wide .env when the
+        # default provider is oci_sdk. Translate only the well-known OCI host.
+        if url.startswith("https://inference.generativeai.") and url.endswith(".oci.oraclecloud.com"):
+            url += "/openai/v1"
+        if not url.endswith("/openai/v1"):
+            raise RuntimeError(
+                f"Supervisor OpenAI-compatible base URL must end with /openai/v1, got {url!r}. "
+                "Set base_url under profiles.supervisor to the working endpoint."
+            )
+
+    key = profile.get("api_key") or getattr(settings, "OCI_GENAI_API_KEY", None) or os.getenv("OCI_GENAI_API_KEY")
+    if effective_provider in {"oci_openai", "openai_compatible"} and not key:
+        raise RuntimeError(
+            "Supervisor profile requires OCI_GENAI_API_KEY (or profiles.supervisor.api_key). "
+            "OCI SDK config_file signing is NOT an OpenAI-compatible bearer API key."
+        )
+
+    overrides = {
+        "LLM_PROVIDER": effective_provider,
+        "OCI_GENAI_MODEL": str(profile.get("model") or settings.OCI_GENAI_MODEL),
+        "OCI_GENAI_BASE_URL": url,
+        "LLM_TEMPERATURE": float(profile.get("temperature", settings.LLM_TEMPERATURE)),
+        "LLM_MAX_TOKENS": int(profile.get("max_tokens", settings.LLM_MAX_TOKENS)),
+    }
+    if key:
+        overrides["OCI_GENAI_API_KEY"] = str(key)
+
+    supervisor_settings = settings.model_copy(update=overrides)
+    logger.info(
+        "Supervisor LLM profile loaded path=%s provider=%s model=%s base_url=%s",
+        path, effective_provider, overrides["OCI_GENAI_MODEL"], url,
+    )
+    return create_llm(supervisor_settings)

apps/agent_gateway/config/backends.a2a.example.yaml

@@ -0,0 +1,27 @@
+#default_backend: native
+#backends:
+#  native:
+#    url: http://localhost:8000
+#    description: Agente nativo do framework
+#    keywords: [conta, fatura]
+#    # protocol omitido equivale a native
+#  external:
+#    protocol: a2a
+#    a2a_version: '1.0' # ou '0.3'
+#    url: https://external-agent.example
+#    a2a_rpc_path: /a2a # endpoint definido pelo operador a partir do Agent Card
+#    a2a_agent_card_path: /.well-known/agent-card.json
+#    description: Agente externo A2A
+#    keywords: [suporte]
+#    a2a_headers_env:
+#      Authorization: EXTERNAL_AGENT_AUTHORIZATION # valor completo: Bearer <token>
+
+default_backend: agent_template
+
+backends:
+  agent_template:
+    protocol: a2a
+    a2a_version: "1.0"
+    url: http://localhost:8000
+    a2a_rpc_path: /a2a
+    a2a_agent_card_path: /.well-known/agent-card.json

apps/agent_gateway/config/llm_profiles.yaml

@@ -0,0 +1,93 @@
+profiles:
+  default:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0.2
+    max_tokens: 2048
+  supervisor:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0
+    max_tokens: 700
+  route_continuity:
+    provider: oci_openai
+    model: openai.gpt-4.1-mini
+    temperature: 0
+    max_tokens: 80
+    timeout_seconds: 5
+  processing_interruption_classifier:
+    provider: oci_openai
+    model: openai.gpt-4.1-mini
+    temperature: 0
+    max_tokens: 8
+    timeout_seconds: 5
+  router:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0
+    max_tokens: 768
+  guardrail:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0
+    max_tokens: 600
+  grl:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0
+    max_tokens: 700
+  judge:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0
+    max_tokens: 800
+  rag_rewriter:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0
+    max_tokens: 300
+  rag_compressor:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0
+    max_tokens: 1200
+  rag_generation:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0.1
+    max_tokens: 1800
+  summary_memory:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0.1
+    max_tokens: 1200
+  noc:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0
+    max_tokens: 700
+  billing_agent:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0.2
+  product_agent:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0.2
+  backoffice_agent:
+    provider: oci_openai
+    model: openai.gpt-4.1
+    temperature: 0.2
+  mcp_parameter_extraction:
+    provider: oci_openai
+    model: openai.gpt-4.1-mini
+    temperature: 0
+    max_tokens: 80
+    timeout_seconds: 5
+
+  transaction_parameter_extraction:
+    provider: oci_openai
+    model: openai.gpt-4.1-mini
+    temperature: 0
+    max_tokens: 500
+    timeout_seconds: 8

deploy/oke/load-test/.env.runtime

@@ -1,159 +0,0 @@
-# Generated by prepare_env.py. Do not commit.
-ADB_DSN=oradb23ai_high
-ADB_PASSWORD=Moniquinha19721972
-ADB_TABLE_PREFIX=AGENTFW
-ADB_USER=admin
-ADB_WALLET_LOCATION=/app/wallet
-ADB_WALLET_PASSWORD=Moniquinha1972
-AGENT_PUBSUB_TOPIC=teste_pubsub
-ANALYTICS_PROVIDERS=pubsub
-API_HOST=0.0.0.0
-API_PORT=8000
-APP_ENV=oke-load-test
-APP_NAME=ai-agent-template
-CHECKPOINT_REPOSITORY_PROVIDER=autonomous
-CORS_ORIGINS=http://localhost:5173,http://127.0.0.1:5173
-DEFAULT_CHANNEL=web
-EMBEDDING_PROVIDER=oci
-ENABLE_ANALYTICS=false
-ENABLE_CONVERSATION_SUMMARY_MEMORY=true
-ENABLE_INPUT_GUARDRAILS=true
-ENABLE_JUDGES=true
-ENABLE_LANGFUSE=true
-ENABLE_LANGFUSE_ANALYTICS_PUBLISHER=false
-ENABLE_LANGFUSE_OPENAI_AUTO_INSTRUMENTATION=true
-ENABLE_LLM_ROUTER=true
-ENABLE_LONG_TERM_MEMORY=true
-ENABLE_MCP_TOOLS=false
-ENABLE_OCI_STREAMING=false
-ENABLE_OTEL=false
-ENABLE_OUTPUT_GUARDRAILS=true
-ENABLE_OUTPUT_SUPERVISOR=true
-ENABLE_PARALLEL_GUARDRAILS=true
-ENABLE_REDIS_CACHE=false
-ENABLE_ROUTE_STICKINESS=true
-ENABLE_SUPERVISOR=true
-ENABLE_TEXT_ADAPTER=true
-ENABLE_VOICE_ADAPTER=true
-ENABLE_WHATSAPP_ADAPTER=true
-END_SESSION_MESSAGE='Atendimento encerrado. Obrigado pelo contato.'
-FRAMEWORK_CHANNEL_INPUT_MODE=embedded
-GCP_PROJECT_ID=project-7f829b68-bf43-442c-ad3
-GCP_PUBSUB_TIMEOUT_SECONDS=30
-GCP_PUBSUB_TOPIC=teste_pubsub
-GCP_PUBSUB_TOPIC_PATH=projects/project-7f829b68-bf43-442c-ad3/topics/teste_pubsub
-GOOGLE_APPLICATION_CREDENTIALS=/mnt/d/Dropbox/ORACLE/TIM/FY27/Testes/sa-hoshikawa.json
-GRAPH_STORE_PROVIDER=autonomous
-GUARDRAILS_CONFIG_PATH=./config/guardrails.yaml
-GUARDRAILS_FAIL_FAST=true
-HUMAN_HANDOFF_MESSAGE='Vou encaminhar seu atendimento para uma pessoa.'
-IDENTITY_CONFIG_PATH=./config/identity.yaml
-IMAGE_TAG=loadtest
-JUDGES_CONFIG_PATH=./config/judges.yaml
-K8S_NAMESPACE=agent-load-test
-LANGFUSE_COMPACT_SUPPRESSED_PREFIXES=llm.chat_completion
-LANGFUSE_COMPACT_VISIBLE_EVENT_PREFIXES='AGA.,NOC., IC.'
-LANGFUSE_HOST=http://langfuse-web.langfuse.svc.cluster.local:3000
-LANGFUSE_IGNORED_PATHS=/health,/ready,/metrics
-LANGFUSE_IGNORE_HEALTHCHECKS=true
-LANGFUSE_PUBLIC_KEY=pk-lf-2c82b991104f3f38f41878baecaaffd0
-LANGFUSE_SECRET_KEY=sk-lf-defd21d2c6da3d4a9f301d993236958f9e56d94f9050a00f79067267b78a00a0
-LANGFUSE_TRACE_MODE=compact
-LLM_MAX_TOKENS=2048
-LLM_PROVIDER=oci_sdk
-LLM_TEMPERATURE=0.2
-LLM_TIMEOUT_SECONDS=120
-LOADTEST_AGENT_ID=telecom_contas
-LOADTEST_CPU_LIMIT=2000m
-LOADTEST_CPU_REQUEST=500m
-LOADTEST_DURATION=10m
-LOADTEST_MAX_REPLICAS=30
-LOADTEST_MAX_VUS=1000
-LOADTEST_MEMORY_LIMIT=2Gi
-LOADTEST_MEMORY_REQUEST=768Mi
-LOADTEST_MESSAGE='Explique em uma frase o que é uma fatura de telecom.'
-LOADTEST_MIN_REPLICAS=4
-LOADTEST_RPS=100
-LOADTEST_SCENARIO=unique_sessions
-LOADTEST_TARGET=http://agent-template-backend-lb.agent-load-test.svc.cluster.local:8000
-LOADTEST_TENANT_ID=loadtest
-LOADTEST_VUS=250
-LOG_LEVEL=INFO
-LONG_TERM_MEMORY_AUTO_EXTRACT=true
-LONG_TERM_MEMORY_INJECT_CONTEXT=true
-LONG_TERM_MEMORY_MAX_CONTEXT_ITEMS=20
-LONG_TERM_MEMORY_MIN_CONFIDENCE=0.70
-LONG_TERM_MEMORY_PROVIDER=sqlite
-LONG_TERM_MEMORY_SQLITE_PATH=./data/agent_framework.db
-LONG_TERM_MEMORY_TABLE=agentfw_long_term_memory
-MCP_GATEWAY_ENABLED=false
-MCP_PARAMETER_MAPPING_PATH=./config/mcp_parameter_mapping.yaml
-MCP_SERVERS_CONFIG_PATH=./config/mcp_servers.yaml
-MCP_TOOL_TIMEOUT_SECONDS=30
-MEMORY_CONTEXT_STRATEGY=summary
-MEMORY_HISTORY_LIMIT=80
-MEMORY_INJECT_RECENT_MESSAGES=true
-MEMORY_INJECT_SUMMARY=true
-MEMORY_MAX_SUMMARY_CHARS=6000
-MEMORY_RECENT_MESSAGES_LIMIT=8
-MEMORY_REPOSITORY_PROVIDER=autonomous
-MEMORY_SUMMARY_TRIGGER_MESSAGES=20
-MEMORY_SUMMARY_USE_LLM=true
-MONGODB_DATABASE=agent_platform
-MONGODB_URI='mongodb://user:password@autonomous-private-endpoint:27017/?tls=true'
-OCIR_REPOSITORY_PREFIX=agent-framework-loadtest
-OCI_AUTH_MODE=oke_workload_identity
-OCI_CLI_PROFILE=LATINOAMERICA-SaoPaulo
-OCI_COMPARTMENT_ID=ocid1.compartment.oc1..aaaaaaaaexpiw4a7dio64mkfv2t273s2hgdl6mgfvvyv7tycalnjlvpvfl3q
-OCI_CONFIG_FILE='~/.oci/config'
-OCI_EMBEDDING_MODEL=cohere.embed-multilingual-v3.0
-OCI_GENAI_API_KEY=sk-ph3FgX6iP3fxAQCXb9IpPIDTadkeeYAWntUWhzcWysIM6zsS
-OCI_GENAI_BASE_URL=https://inference.generativeai.us-chicago-1.oci.oraclecloud.com
-OCI_GENAI_MODEL=openai.gpt-4.1
-OCI_GENAI_PROJECT_OCID=''
-OCI_PROFILE=LATINOAMERICA-Chicago
-OCI_REGION=sa-saopaulo-1
-OCI_REGION_KEY=gru
-OCIR_USERNAME=idi1o0a010nx/oracleidentitycloudservice/cristiano.hoshikawa@oracle.com
-OCIR_AUTH_TOKEN="A5XBm>8gvA2oDy;it1lq"
-OCIR_EMAIL=cristiano.hoshikawa@oracle.com
-OCIR_PULL_SECRET_NAME=ocir-secret
-OCI_STREAM_ENDPOINT=''
-OCI_STREAM_OCID=''
-OCI_STREAM_PARTITION_KEY=agent-events
-OCI_TENANCY_NAMESPACE=idi1o0a010nx
-OKE_CLUSTER_OCID=ocid1.cluster.oc1.sa-saopaulo-1.aaaaaaaabpismmmj6kwo4idomzso5fr32xlnbjcm74pfa7bfhc52plqe64oq
-EXISTING_LB_OCID=ocid1.loadbalancer.oc1.sa-saopaulo-1.aaaaaaaaklbmmriyabvxradqkchy5cobelzxj37rn27p2rsfef5uez54jfqa
-BACKEND_SET_NAME=agent-framework-loadtest
-BACKEND_LISTENER_NAME=agent-framework-loadtest
-BACKEND_LISTENER_PORT=8000
-BACKEND_LISTENER_PROTOCOL=HTTP
-BACKEND_HEALTH_PATH=/health
-OTEL_EXPORTER_OTLP_ENDPOINT=''
-OTEL_SERVICE_NAME=ai-agent-template
-OUTPUT_SUPERVISOR_MAX_RETRIES=3
-PROJECT_ID=project-7f829b68-bf43-442c-ad3
-PROMPT_POLICY_PATH=./config/prompt_policy.yaml
-PUBSUB_SEQUENCE_ENABLED=true
-PUBSUB_SEQUENCE_KEY_PREFIX=observer:sequence
-PUBSUB_SEQUENCE_MEMORY_FALLBACK=false
-PUBSUB_SEQUENCE_MONGODB_COLLECTION=observer_event_counters
-PUBSUB_SEQUENCE_MONGODB_DATABASE=agent_platform
-PUBSUB_SEQUENCE_MONGODB_URI='mongodb://user:password@autonomous-private-endpoint:27017/?tls=true'
-PUBSUB_SEQUENCE_PROVIDER=mongodb
-PUBSUB_SEQUENCE_TTL_SECONDS=86400
-PUBSUB_TOPIC_PATH=projects/project-7f829b68-bf43-442c-ad3/topics/teste_pubsub
-RAG_FILE_GLOBS='*.md,*.txt,*.yaml,*.yml,*.json'
-RAG_TOP_K=5
-REDIS_URL=redis://localhost:6379/0
-ROUTE_STICKINESS_CONFIDENCE_THRESHOLD=0.90
-ROUTE_STICKINESS_HISTORY_TURNS=2
-ROUTE_STICKINESS_LLM_PROFILE=route_continuity
-ROUTE_STICKINESS_MAX_TOKENS=80
-ROUTING_CONFIG_PATH=./config/routing.yaml
-ROUTING_MODE=router
-SESSION_REPOSITORY_PROVIDER=autonomous
-SIMULATE_AGA_MISSING_TRANSACTION_ID=true
-TOOLS_CONFIG_PATH=./config/tools.yaml
-USAGE_REPOSITORY_PROVIDER=autonomous
-VECTOR_STORE_PROVIDER=autonomous

docs/A2A_ASSINCRONO.md

@@ -0,0 +1,248 @@
+# Agentes do framework com A2A assíncrono
+
+## O que foi implementado
+
+O agente criado a partir de `templates/agent_template_backend` pode expor o
+protocolo nativo e A2A no mesmo processo. O Agent Gateway também oferece a
+mesma fachada A2A. As duas fachadas usam o SDK oficial e suas operações de
+Tasks/SSE, com interoperabilidade A2A 1.0 e 0.3.
+
+```text
+Consumidor A2A ── Gateway A2A (Task)
+                    ├── agente nativo (/gateway/message)
+                    └── agente framework A2A (Task remota)
+
+Consumidor A2A ── agente framework A2A diretamente
+```
+
+O envio imediato devolve uma Task antes de terminar a operação de negócio.
+A execução continua após a resposta HTTP e também após a desconexão do SSE.
+O consumidor pode consultar o resultado ou se reconectar à tarefa em andamento.
+Os endpoints nativos conservam seu comportamento; `/gateway/message` continua
+aguardando o resultado, e o SSE nativo permanece separado do SSE A2A.
+
+## Habilitar no agente construído com o framework
+
+Na raiz do framework, instale a dependência opcional:
+
+```sh
+python -m pip install -e 'libs/agent_framework[a2a]' -r templates/agent_template_backend/requirements.txt
+```
+
+No ambiente do backend:
+
+```dotenv
+ENABLE_A2A=true
+A2A_TASK_MODE=true
+A2A_PUBLIC_URL=http://localhost:8000
+A2A_TASK_TIMEOUT_SECONDS=240
+```
+
+Inicie o backend com suas configurações normais. Por exemplo, após preparar o
+`.env` e os arquivos de configuração do agente:
+
+```sh
+cd templates/agent_template_backend
+python -m uvicorn app.main:app --host 0.0.0.0 --port 8000
+```
+
+O endereço público real deve ser usado em `A2A_PUBLIC_URL`. A configuração de
+exemplo está em `.env.a2a.example`. Há também uma imagem opcional, construída
+na raiz do framework:
+
+```sh
+docker build -f templates/agent_template_backend/Dockerfile.a2a -t framework-agent-a2a .
+```
+
+Os perfis de `agents.yaml` são publicados como skills no Agent Card:
+`GET /.well-known/agent-card.json`. Use `message.metadata.agent_id` para escolher
+um perfil; sem ele, o backend mantém sua seleção padrão.
+
+## Registrar esse agente no Agent Gateway
+
+```yaml
+backends:
+  pedidos_a2a:
+    protocol: a2a
+    a2a_version: '1.0'
+    url: http://backend-pedidos:8000
+    a2a_rpc_path: /a2a
+    description: Consulta e acompanhamento de pedidos
+    keywords: [pedido, entrega, rastreamento]
+    default_agent_id: pedidos
+```
+
+No gateway, habilite `ENABLE_A2A=true`, `A2A_TASK_MODE=true` e configure seu próprio
+`A2A_PUBLIC_URL`. Para uma execução mais longa, ajuste os limites, por exemplo:
+`A2A_TASK_TIMEOUT_SECONDS=300` e `BACKEND_TIMEOUT_SECONDS=270`, com o timeout do
+backend em 240 segundos. A duração efetiva é limitada pelo menor prazo aplicável.
+
+O gateway devolve sua própria Task imediatamente e consulta a Task remota até
+conclusão ou solicitação de input/autenticação. Eventos de status remoto são
+publicados como `a2a.backend.status`. Os IDs local e remoto podem ser distintos.
+O resultado nativo inclui `metadata.a2a_task_id` e `metadata.a2a_context_id` para
+correlação; o consumidor externo continua acompanhando o ID emitido pelo gateway.
+
+## Envio assíncrono A2A 1.0
+
+Envie um `POST /a2a` com `Content-Type: application/json` e
+`A2A-Version: 1.0`:
+
+```json
+{
+  "jsonrpc": "2.0",
+  "id": "request-1",
+  "method": "SendMessage",
+  "params": {
+    "configuration": {"returnImmediately": true},
+    "message": {
+      "messageId": "message-1",
+      "contextId": "conversation-123",
+      "role": "ROLE_USER",
+      "parts": [{"text": "Consulte meu pedido"}],
+      "metadata": {"agent_id": "pedidos", "tenant_id": "default"}
+    }
+  }
+}
+```
+
+A resposta contém `result.task.id`, `contextId` e um estado inicial como
+`TASK_STATE_SUBMITTED` ou `TASK_STATE_WORKING`. Guarde esse ID e consulte:
+
+```json
+{
+  "jsonrpc": "2.0",
+  "id": "request-2",
+  "method": "GetTask",
+  "params": {"id": "ID_DA_TAREFA"}
+}
+```
+
+O resultado está em `result`. Quando `status.state` for `TASK_STATE_COMPLETED`,
+o conteúdo final estará em `artifacts[].parts`. Há uma parte de texto e uma
+parte `data` com a resposta original do framework.
+
+Sem `returnImmediately: true`, o envio aguarda o estado final ou interrompido,
+retornando uma Task no mesmo formato. Para manter o comportamento da primeira
+entrega A2A, que retornava Message, configure `A2A_TASK_MODE=false`; nesse modo,
+streaming fica desabilitado e não há Tasks locais.
+
+## Streaming e reconexão
+
+Para enviar e acompanhar, use `SendStreamingMessage` no lugar de `SendMessage`
+com os mesmos parâmetros. A resposta tem `Content-Type: text/event-stream`.
+Cada campo `data:` contém uma resposta JSON-RPC. Os eventos contêm:
+
+1. A Task inicial.
+2. Atualizações de status: submitted/working e eventos de progresso.
+3. O artefato de resposta.
+4. O estado completed, failed, canceled, input-required ou auth-required.
+
+O backend do framework publica marcos como `flow.start`, `session.upserted`,
+`workflow.started`, `workflow.completed` e `message.responded`, quando o fluxo
+passa por essas etapas. Eles aparecem em
+`result.statusUpdate.metadata.framework_event`, com os dados correspondentes
+em `framework_payload`. Isso não implica transmissão de tokens do LLM.
+
+Para se reconectar a uma Task em andamento:
+
+```json
+{
+  "jsonrpc": "2.0",
+  "id": "request-3",
+  "method": "SubscribeToTask",
+  "params": {"id": "ID_DA_TAREFA"}
+}
+```
+
+A reconexão entrega o estado atual e os eventos seguintes; não há promessa de
+replay de todos os eventos perdidos. Se a tarefa já terminou, consulte `GetTask`.
+
+## Cancelamento, falhas e continuidade
+
+`CancelTask`, com `params.id`, cancela a execução local em andamento. O SDK
+interrompe a coroutine e registra `TASK_STATE_CANCELED`. Isso não desfaz efeitos
+já confirmados em ferramentas ou sistemas externos. Quando o gateway está
+consultando um agente remoto, parar a tarefa local não garante cancelar a
+operação que já foi aceita pelo serviço remoto.
+
+Exceções e expiração de `A2A_TASK_TIMEOUT_SECONDS` resultam em
+`TASK_STATE_FAILED`, consultável depois, com mensagem genérica. Detalhes internos
+permanecem nos logs do servidor.
+
+Para uma interação que exige nova entrada, um handler pode retornar
+`metadata.a2a_task_state: "input-required"` ou `"auth-required"`. O consumidor
+envia outra mensagem com o mesmo `taskId` e `contextId`. A tarefa é retomada;
+uma tarefa concluída não pode ser reexecutada. O adaptador de saída preserva a
+continuidade de tarefas remotas nesses estados.
+
+Os workflows nativos que não declaram esse estado concluem uma Task por turno;
+a conversa continua com uma nova Task e o mesmo `contextId`, usando as sessões
+e memórias já existentes no framework.
+
+## Equivalentes A2A 0.3
+
+Use `A2A-Version: 0.3`, `role: "user"` e os discriminadores `kind` de 0.3.
+
+| A2A 1.0 | A2A 0.3 |
+| --- | --- |
+| `SendMessage` + `returnImmediately: true` | `message/send` + `blocking: false` |
+| `SendStreamingMessage` | `message/stream` |
+| `GetTask` | `tasks/get` |
+| `SubscribeToTask` | `tasks/resubscribe` |
+| `CancelTask` | `tasks/cancel` |
+
+Em 0.3, `result` contém diretamente a Task, sem o envelope `task`. Os estados
+usam `submitted`, `working`, `completed`, `failed`, etc.
+
+## Armazenamento e autenticação
+
+O padrão usa o TaskStore em memória do SDK. Tarefas e resultados são perdidos
+quando o processo reinicia. Para múltiplos workers/réplicas, use afinidade de
+requisições ao mesmo processo durante a execução ou configure armazenamento e
+coordenação compartilhados. Um TaskStore persistente, sozinho, não torna a
+execução retomável após falha. As funções `install_a2a` e `install_backend_a2a`
+aceitam `task_store` para integração com um armazenamento do SDK.
+
+O backend opcional reutiliza `AGENT_AUTH_*`; o gateway usa
+`AGENT_GATEWAY_AUTH_*`. A identidade autenticada pelo middleware define o dono
+das Tasks, incluindo consultas e cancelamento. Usuários anônimos compartilham
+o escopo anônimo. `tenant_id` em metadata continua sendo dado do pedido, não
+credencial nem isolamento de autorização. Clientes que compartilham uma mesma
+credencial/principal compartilham o escopo de tarefas.
+
+`A2A_SECURITY_SCHEMES` e `A2A_SECURITY` documentam os mecanismos no Agent Card,
+como descrito no guia geral. SSE está habilitado no modo de Tasks;
+notificações push e gRPC continuam fora desta implementação.
+
+## Validação
+
+A suíte inclui envio imediato enquanto o handler permanece bloqueado, polling,
+streaming com progresso e resultado, cancelamento, falhas, timeout, retomada,
+separação entre identidades autenticadas e gateway chamando um backend A2A real
+em aplicação local. Há também testes HTTP em `127.0.0.1` que verificam a chegada
+do primeiro evento antes da conclusão e reconexão após desconexão.
+
+```sh
+python -m pip install pytest pytest-asyncio uvicorn
+PYTHONPATH=libs/agent_framework/src pytest \
+  tests/test_a2a_interoperability.py tests/test_a2a_async_tasks.py \
+  tests/test_a2a_http_streaming.py \
+  tests/unit/test_sse.py tests/unit/test_sse_replay_dedup.py \
+  tests/unit/test_authentication.py tests/unit/test_authentication_policies.py -q
+```
+
+Os testes não acessam OCI nem serviços de negócio externos. Em ambientes sem
+permissão para abrir portas locais, execute o teste HTTP em um ambiente que
+permita bind em loopback.
+
+Especificação oficial: https://a2a-protocol.org/latest/specification/
+SDK oficial: https://github.com/a2aproject/a2a-python
+
+## Frontend pronto para A2A
+
+A interface `apps/agent_frontend_a2a` consome este protocolo diretamente, com
+Tasks, streaming, consulta e cancelamento. Consulte
+[Frontend A2A](../apps/agent_frontend_a2a/README.md) para a demonstração sem OCI
+e os comandos de execução com o backend ou Gateway. O frontend nativo continua
+em `apps/agent_frontend`.

docs/A2A_ASYNC_EN.md

@@ -0,0 +1,180 @@
+# Framework Agents with Asynchronous A2A
+
+## What was implemented
+
+An agent created from `templates/agent_template_backend` can expose both the native protocol and A2A in the same process. The Agent Gateway also provides the same A2A interface. Both interfaces use the official SDK and its Tasks/SSE operations, with A2A 1.0 and 0.3 interoperability.
+
+```text
+A2A consumer ── A2A Gateway (Task)
+                 ├── native agent (/gateway/message)
+                 └── framework A2A agent (remote Task)
+
+A2A consumer ── framework A2A agent directly
+```
+
+Immediate submission returns a Task before the business operation finishes. Execution continues after the HTTP response and even after the SSE connection is closed. The consumer can retrieve the result or reconnect to an in-progress task. Native endpoints retain their behavior: `/gateway/message` still waits for the result, and native SSE remains separate from A2A SSE.
+
+## Enable A2A in an agent built with the framework
+
+From the framework root, install the optional dependency:
+
+```sh
+python -m pip install -e 'libs/agent_framework[a2a]' -r templates/agent_template_backend/requirements.txt
+```
+
+In the backend environment:
+
+```dotenv
+ENABLE_A2A=true
+A2A_TASK_MODE=true
+A2A_PUBLIC_URL=http://localhost:8000
+A2A_TASK_TIMEOUT_SECONDS=240
+```
+
+Start the backend with its usual settings. For example, after preparing the `.env` and agent configuration files:
+
+```sh
+cd templates/agent_template_backend
+python -m uvicorn app.main:app --host 0.0.0.0 --port 8000
+```
+
+Use the actual public address in `A2A_PUBLIC_URL`. A sample configuration is provided in `.env.a2a.example`. An optional image can also be built from the framework root:
+
+```sh
+docker build -f templates/agent_template_backend/Dockerfile.a2a -t framework-agent-a2a .
+```
+
+Profiles defined in `agents.yaml` are published as skills in the Agent Card: `GET /.well-known/agent-card.json`. Use `message.metadata.agent_id` to select a profile; otherwise, the backend retains its default selection.
+
+## Register the agent with the Agent Gateway
+
+```yaml
+backends:
+  pedidos_a2a:
+    protocol: a2a
+    a2a_version: '1.0'
+    url: http://backend-pedidos:8000
+    a2a_rpc_path: /a2a
+    description: Order inquiry and tracking
+    keywords: [pedido, entrega, rastreamento]
+    default_agent_id: pedidos
+```
+
+On the gateway, enable `ENABLE_A2A=true` and `A2A_TASK_MODE=true`, and configure its own `A2A_PUBLIC_URL`. For longer-running operations, adjust the limits; for example, use `A2A_TASK_TIMEOUT_SECONDS=300` and `BACKEND_TIMEOUT_SECONDS=270`, with the backend timeout set to 240 seconds. The effective duration is limited by the shortest applicable deadline.
+
+The gateway immediately returns its own Task and polls the remote Task until completion or until input/authentication is required. Remote status events are published as `a2a.backend.status`. Local and remote IDs may differ. The native result includes `metadata.a2a_task_id` and `metadata.a2a_context_id` for correlation; external consumers continue tracking the ID issued by the gateway.
+
+## Asynchronous A2A 1.0 submission
+
+Send a `POST /a2a` request with `Content-Type: application/json` and `A2A-Version: 1.0`:
+
+```json
+{
+  "jsonrpc": "2.0",
+  "id": "request-1",
+  "method": "SendMessage",
+  "params": {
+    "configuration": {"returnImmediately": true},
+    "message": {
+      "messageId": "message-1",
+      "contextId": "conversation-123",
+      "role": "ROLE_USER",
+      "parts": [{"text": "Check my order"}],
+      "metadata": {"agent_id": "pedidos", "tenant_id": "default"}
+    }
+  }
+}
+```
+
+The response contains `result.task.id`, `contextId`, and an initial state such as `TASK_STATE_SUBMITTED` or `TASK_STATE_WORKING`. Save that ID and query the task:
+
+```json
+{
+  "jsonrpc": "2.0",
+  "id": "request-2",
+  "method": "GetTask",
+  "params": {"id": "TASK_ID"}
+}
+```
+
+The result is in `result`. When `status.state` is `TASK_STATE_COMPLETED`, the final content is in `artifacts[].parts`. It includes a text part and a `data` part containing the original framework response.
+
+Without `returnImmediately: true`, submission waits for a final or interrupted state and returns a Task in the same format. To preserve the behavior of the initial A2A release, which returned a Message, set `A2A_TASK_MODE=false`. In that mode, streaming is disabled and there are no local Tasks.
+
+## Streaming and reconnection
+
+To submit and follow a task, use `SendStreamingMessage` instead of `SendMessage`, with the same parameters. The response uses `Content-Type: text/event-stream`. Each `data:` field contains a JSON-RPC response. Events include:
+
+1. The initial Task.
+2. Status updates (submitted/working) and progress events.
+3. The response artifact.
+4. A completed, failed, canceled, input-required, or auth-required state.
+
+The framework backend publishes milestones such as `flow.start`, `session.upserted`, `workflow.started`, `workflow.completed`, and `message.responded` when the flow reaches those stages. They appear in `result.statusUpdate.metadata.framework_event`, with corresponding data in `framework_payload`. This does not imply LLM token streaming.
+
+To reconnect to an in-progress Task:
+
+```json
+{
+  "jsonrpc": "2.0",
+  "id": "request-3",
+  "method": "SubscribeToTask",
+  "params": {"id": "TASK_ID"}
+}
+```
+
+Reconnection delivers the current state and subsequent events; replay of all missed events is not guaranteed. If the task has already finished, use `GetTask`.
+
+## Cancellation, failures, and continuity
+
+`CancelTask`, with `params.id`, cancels the ongoing local execution. The SDK interrupts the coroutine and records `TASK_STATE_CANCELED`. This does not undo effects already committed in tools or external systems. When the gateway is polling a remote agent, stopping the local task does not guarantee cancellation of an operation already accepted by the remote service.
+
+Exceptions and expiration of `A2A_TASK_TIMEOUT_SECONDS` result in `TASK_STATE_FAILED`, which can be queried later and carries a generic message. Internal details remain in server logs.
+
+For an interaction requiring further input, a handler can return `metadata.a2a_task_state: "input-required"` or `"auth-required"`. The consumer sends another message with the same `taskId` and `contextId`. The task is resumed; a completed task cannot be re-executed. The output adapter preserves continuity for remote tasks in these states.
+
+Native workflows that do not declare such a state complete one Task per turn. The conversation continues with a new Task and the same `contextId`, using the framework's existing sessions and memories.
+
+## A2A 0.3 equivalents
+
+Use `A2A-Version: 0.3`, `role: "user"`, and the 0.3 `kind` discriminators.
+
+| A2A 1.0 | A2A 0.3 |
+| --- | --- |
+| `SendMessage` + `returnImmediately: true` | `message/send` + `blocking: false` |
+| `SendStreamingMessage` | `message/stream` |
+| `GetTask` | `tasks/get` |
+| `SubscribeToTask` | `tasks/resubscribe` |
+| `CancelTask` | `tasks/cancel` |
+
+In 0.3, `result` directly contains the Task without the `task` envelope. States use `submitted`, `working`, `completed`, `failed`, etc.
+
+## Storage and authentication
+
+By default, the SDK uses an in-memory TaskStore. Tasks and results are lost when the process restarts. With multiple workers/replicas, use request affinity to the same process during execution or configure shared storage and coordination. A persistent TaskStore alone does not make execution resumable after a failure. The `install_a2a` and `install_backend_a2a` functions accept `task_store` for integration with SDK storage.
+
+The optional backend reuses `AGENT_AUTH_*`; the gateway uses `AGENT_GATEWAY_AUTH_*`. The identity authenticated by middleware determines Task ownership, including retrieval and cancellation. Anonymous users share the anonymous scope. `tenant_id` in metadata remains request data, not a credential or authorization isolation mechanism. Clients sharing the same credential/principal also share a task scope.
+
+`A2A_SECURITY_SCHEMES` and `A2A_SECURITY` document mechanisms in the Agent Card, as described in the general guide. SSE is enabled in Task mode; push notifications and gRPC are outside the scope of this implementation.
+
+## Validation
+
+The test suite covers immediate submission while the handler remains blocked, polling, streaming with progress and results, cancellation, failures, timeouts, resumption, isolation between authenticated identities, and a gateway calling a real A2A backend in a local application. HTTP tests on `127.0.0.1` also verify that the first event arrives before completion and that reconnection works after disconnection.
+
+```sh
+python -m pip install pytest pytest-asyncio uvicorn
+PYTHONPATH=libs/agent_framework/src pytest \
+  tests/test_a2a_interoperability.py tests/test_a2a_async_tasks.py \
+  tests/test_a2a_http_streaming.py \
+  tests/unit/test_sse.py tests/unit/test_sse_replay_dedup.py \
+  tests/unit/test_authentication.py tests/unit/test_authentication_policies.py -q
+```
+
+The tests do not access OCI or external business services. In environments where binding to local ports is not permitted, run the HTTP test in an environment that allows binding to loopback.
+
+Official specification: https://a2a-protocol.org/latest/specification/  
+Official SDK: https://github.com/a2aproject/a2a-python
+
+## A2A-ready frontend
+
+The `apps/agent_frontend_a2a` interface consumes this protocol directly, supporting Tasks, streaming, retrieval, and cancellation. See [A2A Frontend](../apps/agent_frontend_a2a/README.md) for the demonstration without OCI and startup commands using the backend or Gateway. The native frontend remains in `apps/agent_frontend`.

docs/A2A_IMPLEMENTACAO.md

@@ -0,0 +1,145 @@
+# Integração A2A no Agent Framework OCI
+
+## Decisão de arquitetura
+
+O Agent Gateway existente passa a ter uma fachada A2A opcional. Não é necessário
+manter um segundo roteador, duplicar o supervisor ou migrar os agentes atuais.
+
+```text
+Consumidor atual ── /gateway/message ─┐
+                                    ├─ Agent Gateway / supervisor ─┬─ backend native
+Consumidor A2A ──── /a2a ─────────────┘                             └─ backend a2a
+```
+
+O backend nativo continua recebendo `{channel, payload, agent_id, tenant_id}`.
+A ausência de `protocol` no catálogo mantém `native`. A2A fica desativado por
+padrão; o SDK só é importado ao habilitar a fachada ou selecionar um backend A2A.
+Não foi alterado o runtime dos agentes, os workflows, os canais ou os MCPs.
+
+A entrada usa A2A **1.0 / JSON-RPC**, SDK oficial `a2a-sdk==1.2.2`, com adaptação
+0.3 disponibilizada pelo próprio SDK. Agentes externos podem ser registrados
+com `a2a_version: '1.0'` ou `'0.3'`; não há detecção heurística de versão.
+
+## Ativação local
+
+Na raiz do framework:
+
+```sh
+python -m pip install -e 'libs/agent_framework[a2a]' -r apps/agent_gateway/requirements.txt
+export ENABLE_A2A=true
+export A2A_PUBLIC_URL=https://meu-gateway.example
+export BACKENDS_CONFIG_PATH="$PWD/apps/agent_gateway/config/backends.a2a.example.yaml"
+python -m uvicorn app.main:app --app-dir apps/agent_gateway --env-file .env --port 8010
+```
+
+Edite primeiro as URLs do catálogo de exemplo. Para agentes externos, informe o
+endpoint JSON-RPC anunciado no Agent Card como `url` + `a2a_rpc_path`.
+Credenciais ficam em variáveis de ambiente referenciadas por `a2a_headers_env`.
+Os valores não são publicados em `/backends`.
+
+A imagem opcional pode ser construída a partir da raiz:
+
+```sh
+docker build -f apps/agent_gateway/Dockerfile.a2a -t agent-gateway-a2a .
+```
+
+Defina `A2A_PUBLIC_URL` no ambiente do container e monte o catálogo de backends.
+A imagem e os comandos anteriores de implantação continuam disponíveis.
+
+## Descoberta e chamada
+
+O Agent Card fica em `GET /.well-known/agent-card.json`; as skills são derivadas
+dos backends habilitados. As mensagens chegam em `POST /a2a`.
+
+Exemplo A2A 1.0, com cabeçalho `A2A-Version: 1.0`:
+
+```json
+{
+  "jsonrpc": "2.0",
+  "id": "request-1",
+  "method": "SendMessage",
+  "params": {
+    "message": {
+      "messageId": "message-1",
+      "contextId": "conversation-123",
+      "role": "ROLE_USER",
+      "parts": [
+        {"text": "Quero consultar minha fatura"},
+        {"data": {"customer_id": "123"}}
+      ],
+      "metadata": {"tenant_id": "default"}
+    }
+  }
+}
+```
+
+No modo de tarefas (padrão), a resposta fica em `result.task`. O resultado de
+negócio está em `artifacts[].parts`, com texto e uma parte `data` contendo a
+resposta do framework. Com `A2A_TASK_MODE=false`, permanece em `result.message`. Reutilize `contextId` nos próximos turnos. No modo 0.3,
+use o cabeçalho `A2A-Version: 0.3`, método `message/send`, role `user` e os
+campos `kind` da especificação 0.3; o resultado é uma Task diretamente
+(ou Message com `A2A_TASK_MODE=false`).
+
+Partes `data` devem conter objetos e são incorporadas ao payload de negócio.
+A identidade da sessão e o texto do protocolo prevalecem sobre esses dados.
+Arquivos recebidos pela fachada são rejeitados explicitamente.
+
+## Conversas, tarefas e limites
+
+- A fachada cria **Tasks A2A** e permite envio imediato, consulta, cancelamento
+  e streaming SSE. `A2A_TASK_MODE=false` mantém o modo anterior de Messages.
+  Consulte [A2A assíncrono](A2A_ASSINCRONO.md) para os exemplos completos.
+- O adaptador de saída aceita Message ou Task; consulta `GetTask`/`tasks/get`
+  enquanto o agente está trabalhando, dentro de um prazo total configurável
+  por `BACKEND_TIMEOUT_SECONDS`. Não repete automaticamente operações de envio.
+- Tarefas que pedem input/autenticação preservam estado e identidade para o
+  próximo turno. Tarefas concluídas não são reutilizadas.
+- A continuidade remota é guardada no processo, por tenant/backend/sessão,
+  com validade de uma hora e limite de 10 mil entradas. Em múltiplas réplicas,
+  use afinidade de sessão ou implemente armazenamento compartilhado antes de
+  depender dessa continuidade. Reiniciar o processo limpa esse cache.
+- `metadata.a2a_result` preserva o resultado remoto, incluindo dados/artefatos;
+  `a2a_context_id`, `a2a_task_id` e `a2a_state` permitem inspeção e continuidade
+  explícita no payload quando necessário.
+- Streaming SSE está disponível no modo de tarefas. Push notifications e gRPC
+  não são oferecidos. O proxy de eventos nativo não é utilizado para agentes
+  A2A: use `SendStreamingMessage`/`SubscribeToTask` e seus equivalentes 0.3.
+- O template `templates/agent_template_backend` também expõe A2A diretamente
+  quando `ENABLE_A2A=true`, sem passar pelo gateway global.
+
+## Autenticação
+
+As rotas A2A compartilham o middleware de autenticação já instalado no gateway
+(`AGENT_GATEWAY_AUTH_*`). Habilitar A2A não habilita autenticação automaticamente.
+O tenant em metadata mantém a semântica atual do framework: é dado do pedido,
+não uma autorização concedida pelo protocolo. Políticas de acesso a tenants
+continuam sendo responsabilidade da configuração de autenticação/autorização.
+
+Quando houver autenticação, declare-a também no Agent Card, por exemplo:
+
+```sh
+export A2A_SECURITY_SCHEMES='{"gatewayBearer":{"httpAuthSecurityScheme":{"scheme":"bearer"}}}'
+export A2A_SECURITY='[{"gatewayBearer":[]}]'
+```
+
+Essas declarações documentam o mecanismo; quem verifica as credenciais é o
+middleware. Se desejar descoberta pública, adicione somente o caminho do Agent
+Card à configuração de caminhos públicos; `/a2a` permanece protegido.
+
+## Validação
+
+Testes em `tests/test_a2a_interoperability.py` cobrem os dois formatos de entrada,
+o contrato nativo, os dois formatos de saída, polling, erros JSON-RPC,
+continuidade de tarefas, separação entre tenants e limite de tempo.
+
+```sh
+PYTHONPATH=libs/agent_framework/src pytest tests/test_a2a_interoperability.py -q
+```
+
+Validação com transportes HTTP simulados e aplicação FastAPI local; não substitui
+um teste de integração com os serviços OCI e agentes externos da implantação.
+
+Referências oficiais:
+- https://a2a-protocol.org/latest/specification/
+- https://a2a-protocol.org/v0.3.0/specification/
+- https://github.com/a2aproject/a2a-python

docs/A2A_IMPLEMENTATION_EN.md

@@ -0,0 +1,105 @@
+# A2A Integration in Agent Framework OCI
+
+## Architecture decision
+
+The existing Agent Gateway gains an optional A2A interface. There is no need to maintain a second router, duplicate the supervisor, or migrate existing agents.
+
+```text
+Existing consumer ── /gateway/message ─┐
+                                       ├─ Agent Gateway / supervisor ─┬─ native backend
+A2A consumer ─────── /a2a ─────────────┘                              └─ A2A backend
+```
+
+The native backend continues receiving `{channel, payload, agent_id, tenant_id}`. Omitting `protocol` from the catalog preserves `native`. A2A is disabled by default; the SDK is imported only when the interface is enabled or an A2A backend is selected. The agent runtime, workflows, channels, and MCPs were not changed.
+
+The inbound interface uses A2A **1.0 / JSON-RPC**, the official `a2a-sdk==1.2.2`, with a 0.3 adapter provided by the SDK itself. External agents can be registered with `a2a_version: '1.0'` or `'0.3'`; there is no heuristic version detection.
+
+## Local setup
+
+From the framework root:
+
+```sh
+python -m pip install -e 'libs/agent_framework[a2a]' -r apps/agent_gateway/requirements.txt
+export ENABLE_A2A=true
+export A2A_PUBLIC_URL=https://meu-gateway.example
+export BACKENDS_CONFIG_PATH="$PWD/apps/agent_gateway/config/backends.a2a.example.yaml"
+python -m uvicorn app.main:app --app-dir apps/agent_gateway --env-file .env --port 8010
+```
+
+First, edit the URLs in the sample catalog. For external agents, specify the JSON-RPC endpoint advertised in the Agent Card as `url` + `a2a_rpc_path`. Credentials are stored in environment variables referenced by `a2a_headers_env`. Their values are not exposed through `/backends`.
+
+An optional image can be built from the framework root:
+
+```sh
+docker build -f apps/agent_gateway/Dockerfile.a2a -t agent-gateway-a2a .
+```
+
+Set `A2A_PUBLIC_URL` in the container environment and mount the backend catalog. The existing deployment images and commands remain available.
+
+## Discovery and invocation
+
+The Agent Card is available at `GET /.well-known/agent-card.json`; skills are derived from enabled backends. Messages arrive at `POST /a2a`.
+
+A2A 1.0 example, with the `A2A-Version: 1.0` header:
+
+```json
+{
+  "jsonrpc": "2.0",
+  "id": "request-1",
+  "method": "SendMessage",
+  "params": {
+    "message": {
+      "messageId": "message-1",
+      "contextId": "conversation-123",
+      "role": "ROLE_USER",
+      "parts": [
+        {"text": "I want to check my bill"},
+        {"data": {"customer_id": "123"}}
+      ],
+      "metadata": {"tenant_id": "default"}
+    }
+  }
+}
+```
+
+In Task mode (the default), the response is in `result.task`. The business result is in `artifacts[].parts`, with text and a `data` part containing the framework response. With `A2A_TASK_MODE=false`, it remains in `result.message`. Reuse `contextId` on subsequent turns. In 0.3 mode, use the `A2A-Version: 0.3` header, the `message/send` method, the `user` role, and the `kind` fields from the 0.3 specification; the result is a Task directly (or a Message with `A2A_TASK_MODE=false`).
+
+`data` parts must contain objects and are incorporated into the business payload. Session identity and protocol text take precedence over those data. Files received through the interface are explicitly rejected.
+
+## Conversations, tasks, and limits
+
+- The interface creates **A2A Tasks** and supports immediate submission, retrieval, cancellation, and SSE streaming. `A2A_TASK_MODE=false` preserves the previous Message mode. See [Asynchronous A2A](A2A_ASSINCRONO.md) for complete examples.
+- The output adapter accepts a Message or a Task; it polls `GetTask`/`tasks/get` while the agent is working, within an overall timeout configured by `BACKEND_TIMEOUT_SECONDS`. It does not automatically retry submission operations.
+- Tasks requiring input/authentication preserve state and identity for the next turn. Completed tasks are not reused.
+- Remote continuity is stored in-process by tenant/backend/session, with a one-hour lifetime and a limit of 10,000 entries. With multiple replicas, use session affinity or implement shared storage before relying on this continuity. Process restarts clear the cache.
+- `metadata.a2a_result` preserves the remote result, including data/artifacts; `a2a_context_id`, `a2a_task_id`, and `a2a_state` allow explicit inspection and continuity in the payload when needed.
+- SSE streaming is available in Task mode. Push notifications and gRPC are not provided. The native event proxy is not used for A2A agents: use `SendStreamingMessage`/`SubscribeToTask` and their 0.3 equivalents.
+- The `templates/agent_template_backend` template also exposes A2A directly when `ENABLE_A2A=true`, without going through the global gateway.
+
+## Authentication
+
+A2A routes share the authentication middleware already installed in the gateway (`AGENT_GATEWAY_AUTH_*`). Enabling A2A does not automatically enable authentication. The tenant in metadata retains the framework's current semantics: it is request data, not an authorization granted by the protocol. Tenant access policies remain the responsibility of the authentication/authorization configuration.
+
+When authentication is enabled, also declare it in the Agent Card, for example:
+
+```sh
+export A2A_SECURITY_SCHEMES='{"gatewayBearer":{"httpAuthSecurityScheme":{"scheme":"bearer"}}}'
+export A2A_SECURITY='[{"gatewayBearer":[]}]'
+```
+
+These declarations document the mechanism; the middleware validates credentials. If public discovery is desired, add only the Agent Card path to the public-path configuration; `/a2a` remains protected.
+
+## Validation
+
+Tests in `tests/test_a2a_interoperability.py` cover both input formats, the native contract, both output formats, polling, JSON-RPC errors, task continuity, tenant isolation, and timeouts.
+
+```sh
+PYTHONPATH=libs/agent_framework/src pytest tests/test_a2a_interoperability.py -q
+```
+
+Validation uses simulated HTTP transports and a local FastAPI application; it does not replace integration testing with OCI services and external agents in the deployment environment.
+
+Official references:
+- https://a2a-protocol.org/latest/specification/
+- https://a2a-protocol.org/v0.3.0/specification/
+- https://github.com/a2aproject/a2a-python

examples/a2a_frontend_demo.py

@@ -0,0 +1,43 @@
+"""Local frontend demo, using the actual framework A2A adapter and SDK.
+
+From the framework root:
+  pip install -e 'libs/agent_framework[a2a]' uvicorn
+  uvicorn examples.a2a_frontend_demo:app --host 127.0.0.1 --port 8000
+
+This demo simulates business processing; it never calls OCI or an external tool.
+"""
+import asyncio
+import os
+from pathlib import Path
+from types import SimpleNamespace
+
+from fastapi import FastAPI
+from fastapi.staticfiles import StaticFiles
+from agent_framework.a2a.server import install_backend_a2a
+from agent_framework.a2a.progress import report_progress
+
+app = FastAPI(title='Agent Framework · A2A Frontend Demo')
+
+
+async def demo_agent(request):
+    text = request['payload']['message']
+    await report_progress('workflow.started', {'agent': request.get('agent_id') or 'demo'})
+    await asyncio.sleep(1)
+    if text.strip().lower() == '/erro':
+        raise RuntimeError('Simulated demo error')
+    if text.strip().lower() == '/lento':
+        await asyncio.sleep(20)
+    await report_progress('workflow.completed', {'agent': 'demo'})
+    if text.strip().lower() == '/confirmar':
+        return {'text': 'Deseja continuar com a demonstração? Responda sim para prosseguir.',
+                'metadata': {'a2a_task_state': 'input-required', 'agent_id': 'demo'}}
+    return {'text': f'Recebi sua mensagem: “{text}”.\n\nA tarefa A2A foi concluída. Este exemplo usa o adaptador real do framework, com processamento de negócio simulado.',
+            'session_id': request['session_id'], 'metadata': {'agent_id': 'demo'}}
+
+
+profiles = SimpleNamespace(list_profiles=lambda: [SimpleNamespace(
+    agent_id='demo', name='Agente de demonstração', description='Exemplo local de agente A2A assíncrono')])
+settings = SimpleNamespace(A2A_PUBLIC_URL=os.getenv('A2A_PUBLIC_URL', 'http://localhost:8000'),
+    A2A_TASK_MODE=True, A2A_TASK_TIMEOUT_SECONDS=60, A2A_SECURITY_SCHEMES={}, A2A_SECURITY=[], APP_NAME='Agente de demonstração')
+install_backend_a2a(app, demo_agent, profiles, settings)
+app.mount('/', StaticFiles(directory=Path(__file__).resolve().parents[1] / 'apps' / 'agent_frontend_a2a', html=True), name='frontend')

libs/agent_framework/pyproject.toml

@@ -27,6 +27,9 @@ dependencies = [
   "PyJWT[crypto]>=2.9.0"
 ]
 
+[project.optional-dependencies]
+a2a = ["a2a-sdk[http-server]==1.2.2"]
+
 [tool.setuptools.packages.find]
 where = ["src"]
 
@@ -35,4 +38,4 @@ requires = ["setuptools>=80", "wheel>=0.45"]
 build-backend = "setuptools.build_meta"
 
 [tool.setuptools.package-data]
-"agent_framework" = ["config/*.yaml", "guardrails/calibrated/capabilities/*.yaml"]
+"agent_framework" = ["locales/*.json", "config/*.yaml", "guardrails/calibrated/capabilities/*.yaml"]

libs/agent_framework/src/agent_framework.egg-info/PKG-INFO

@@ -24,3 +24,5 @@ Requires-Dist: motor>=3.6.0
 Requires-Dist: google-cloud-pubsub>=2.28.0
 Requires-Dist: mcp>=1.9.0
 Requires-Dist: PyJWT[crypto]>=2.9.0
+Provides-Extra: a2a
+Requires-Dist: a2a-sdk[http-server]==1.2.2; extra == "a2a"

libs/agent_framework/src/agent_framework.egg-info/SOURCES.txt

@@ -3,6 +3,7 @@ src/agent_framework/__init__.py
 src/agent_framework/extensions.py
 src/agent_framework/gateway_policy_context.py
 src/agent_framework/idempotency.py
+src/agent_framework/localization.py
 src/agent_framework/observer.py
 src/agent_framework/runtime_mcp_gateway_adapter.py
 src/agent_framework.egg-info/PKG-INFO
@@ -10,6 +11,10 @@ src/agent_framework.egg-info/SOURCES.txt
 src/agent_framework.egg-info/dependency_links.txt
 src/agent_framework.egg-info/requires.txt
 src/agent_framework.egg-info/top_level.txt
+src/agent_framework/a2a/__init__.py
+src/agent_framework/a2a/client.py
+src/agent_framework/a2a/progress.py
+src/agent_framework/a2a/server.py
 src/agent_framework/analytics/__init__.py
 src/agent_framework/analytics/composite_publisher.py
 src/agent_framework/analytics/event_builder.py
@@ -39,6 +44,10 @@ src/agent_framework/config/__init__.py
 src/agent_framework/config/agent_registry.py
 src/agent_framework/config/observability_mapping.yaml
 src/agent_framework/config/settings.py
+src/agent_framework/decisions/__init__.py
+src/agent_framework/decisions/client.py
+src/agent_framework/decisions/adapters/__init__.py
+src/agent_framework/decisions/adapters/typesafe.py
 src/agent_framework/events/__init__.py
 src/agent_framework/events/oci_streaming.py
 src/agent_framework/gateways/__init__.py
@@ -138,6 +147,8 @@ src/agent_framework/llm/profile_resolver.py
 src/agent_framework/llm/providers.py
 src/agent_framework/llm/structured_output.py
 src/agent_framework/llm/types.py
+src/agent_framework/locales/en.json
+src/agent_framework/locales/pt-BR.json
 src/agent_framework/mcp/__init__.py
 src/agent_framework/mcp/client.py
 src/agent_framework/mcp/models.py
@@ -174,6 +185,7 @@ src/agent_framework/observability/noc_events.py
 src/agent_framework/observability/noc_otel.py
 src/agent_framework/observability/observer.py
 src/agent_framework/observability/otel.py
+src/agent_framework/observability/severity_policy.py
 src/agent_framework/observability/streaming_events.py
 src/agent_framework/observability/streaming_exporter.py
 src/agent_framework/observability/telemetry.py

libs/agent_framework/src/agent_framework.egg-info/requires.txt

@@ -19,3 +19,6 @@ motor>=3.6.0
 google-cloud-pubsub>=2.28.0
 mcp>=1.9.0
 PyJWT[crypto]>=2.9.0
+
+[a2a]
+a2a-sdk[http-server]==1.2.2

libs/agent_framework/src/agent_framework/a2a/__init__.py

@@ -0,0 +1 @@
+"""Optional A2A interoperability; imported only when enabled."""

libs/agent_framework/src/agent_framework/a2a/client.py

@@ -0,0 +1,108 @@
+"""A2A JSON-RPC adapter. URLs and credential references are operator configured."""
+import asyncio
+import json
+import os
+from uuid import uuid4, uuid5, NAMESPACE_URL
+
+import httpx
+from google.protobuf.json_format import ParseDict, MessageToDict
+from a2a.types import SendMessageResponse, Task
+
+from .progress import report_progress
+
+
+def _normalize_legacy(result):
+    # Preserve the original wire response separately; normalize only presentation.
+    if result.get('kind') == 'message':
+        return {'message': result}
+    if result.get('kind') == 'task':
+        return {'task': result}
+    raise ValueError('Expected an A2A Message or Task')
+
+
+async def call_a2a(backend, request, timeout_seconds):
+    payload = request.get('payload') or {}
+    session = payload.get('session_id') or str(uuid4())
+    tenant = request.get('tenant_id') or 'default'
+    context_id = str(uuid5(NAMESPACE_URL, json.dumps([tenant, backend.backend_id, session])))
+    legacy = backend.a2a_version == '0.3'
+    text = str(payload.get('message') or payload.get('text') or payload.get('transcript') or payload.get('content') or '')
+    parts = [{'text': text}, {'data': payload}]
+    message = {'role': 'ROLE_USER', 'messageId': str(uuid4()),
+               'contextId': payload.get('a2a_context_id') or context_id, 'parts': parts,
+               'metadata': {'tenant_id': tenant, 'agent_id': request.get('agent_id') or backend.default_agent_id}}
+    if payload.get('a2a_task_id'):
+        message['taskId'] = payload['a2a_task_id']
+    if legacy:
+        message.update(kind='message', role='user')
+        for part in parts:
+            part['kind'] = 'text' if 'text' in part else 'data'
+    headers = {'A2A-Version': backend.a2a_version}
+    for header, env_name in backend.a2a_headers_env.items():
+        value = os.environ.get(env_name)
+        if not value:
+            raise ValueError(f'Missing credential environment variable: {env_name}')
+        headers[header] = value
+    endpoint = backend.base_url + backend.a2a_rpc_path
+
+    async with asyncio.timeout(timeout_seconds):
+        async with httpx.AsyncClient(timeout=timeout_seconds, headers=headers) as client:
+            async def rpc(method, params):
+                request_id = str(uuid4())
+                response = await client.post(endpoint, json={
+                    'jsonrpc': '2.0', 'id': request_id, 'method': method, 'params': params,
+                })
+                response.raise_for_status()
+                data = response.json()
+                if data.get('jsonrpc') != '2.0' or data.get('id') != request_id:
+                    raise ValueError('Invalid A2A JSON-RPC response')
+                if 'error' in data:
+                    raise RuntimeError(f'A2A error {data["error"].get("code")}')
+                result = data.get('result')
+                if not isinstance(result, dict):
+                    raise ValueError('Missing A2A result')
+                return result
+
+            raw = await rpc('message/send' if legacy else 'SendMessage', {
+                'message': message,
+                'configuration': {'blocking': False} if legacy else {'returnImmediately': True},
+            })
+            result = _normalize_legacy(raw) if legacy else raw
+            if not legacy:
+                result = MessageToDict(ParseDict(result, SendMessageResponse()))
+            while 'task' in result:
+                task = result['task']
+                state = task.get('status', {}).get('state', '')
+                await report_progress('a2a.backend.status', {
+                    'backend_id': backend.backend_id, 'task_id': task.get('id'), 'state': state,
+                })
+                if state not in {'submitted', 'working', 'TASK_STATE_SUBMITTED', 'TASK_STATE_WORKING'}:
+                    break
+                await asyncio.sleep(backend.a2a_poll_interval_seconds)
+                raw = await rpc('tasks/get' if legacy else 'GetTask', {'id': task['id']})
+                result = {'task': raw if legacy else MessageToDict(ParseDict(raw, Task()))}
+
+    task = result.get('task')
+    if task:
+        state, task_id = task['status']['state'], task['id']
+        if state in {'failed', 'canceled', 'rejected', 'unknown', 'TASK_STATE_FAILED', 'TASK_STATE_CANCELED', 'TASK_STATE_REJECTED', 'TASK_STATE_UNSPECIFIED'}:
+            raise RuntimeError(f'A2A task ended with state: {state}')
+        output_parts = [p for a in task.get('artifacts', []) for p in a.get('parts', [])]
+        if not output_parts and task['status'].get('message'):
+            output_parts += task['status']['message'].get('parts', [])
+        if not output_parts:
+            messages = [m for m in task.get('history', []) if m.get('role') in {'agent', 'ROLE_AGENT'}]
+            output_parts = messages[-1].get('parts', []) if messages else []
+        remote_context = task.get('contextId')
+    elif result.get('message'):
+        reply = result['message']
+        output_parts = reply.get('parts', [])
+        state, task_id, remote_context = None, None, reply.get('contextId')
+    else:
+        raise ValueError('Expected an A2A Message or Task')
+    return {
+        'channel': request.get('channel', 'web'), 'session_id': session,
+        'text': '\n'.join(p['text'] for p in output_parts if 'text' in p),
+        'metadata': {'a2a_context_id': remote_context, 'a2a_task_id': task_id,
+                     'a2a_state': state, 'a2a_result': raw},
+    }

libs/agent_framework/src/agent_framework/a2a/progress.py

@@ -0,0 +1,10 @@
+"""Task-local progress hook; safe to import without the optional A2A SDK."""
+from contextvars import ContextVar
+
+_progress_callback = ContextVar('framework_a2a_progress', default=None)
+
+
+async def report_progress(event: str, payload: dict | None = None):
+    callback = _progress_callback.get()
+    if callback is not None:
+        await callback(event, payload or {})

libs/agent_framework/src/agent_framework/a2a/server.py

@@ -0,0 +1,181 @@
+"""A2A 1.0 Tasks and SSE with SDK-managed 0.3 compatibility."""
+import asyncio
+import logging
+from uuid import uuid4
+
+from google.protobuf.json_format import MessageToDict, ParseDict
+from a2a.auth.user import User
+from a2a.server.agent_execution import AgentExecutor
+from a2a.server.routes import (
+    DefaultServerCallContextBuilder, add_a2a_routes_to_fastapi,
+    create_agent_card_routes, create_jsonrpc_routes,
+)
+from a2a.server.request_handlers import DefaultRequestHandler
+from a2a.server.tasks import InMemoryTaskStore, TaskUpdater
+from a2a.types import AgentCard, Message, Part, Role, Task, TaskState, TaskStatus
+from a2a.utils.errors import InvalidParamsError, TaskNotCancelableError, UnsupportedOperationError
+
+from .progress import _progress_callback
+
+logger = logging.getLogger(__name__)
+
+
+class _PrincipalUser(User):
+    def __init__(self, principal):
+        self.principal = principal
+
+    @property
+    def is_authenticated(self):
+        return True
+
+    @property
+    def user_name(self):
+        # Use middleware identity, never client-supplied message metadata.
+        return f'{self.principal.scheme}:{self.principal.subject}'
+
+
+class FrameworkCallContextBuilder(DefaultServerCallContextBuilder):
+    def build_user(self, request):
+        principal = getattr(request.state, 'auth_principal', None)
+        return _PrincipalUser(principal) if principal else super().build_user(request)
+
+
+def _response_message(response, context_id, task_id=None):
+    data = {
+        'messageId': str(uuid4()), 'role': 'ROLE_AGENT', 'contextId': context_id,
+        'parts': [{'text': str(response.get('text') or response.get('speak') or '')},
+                  {'data': response}],
+    }
+    if task_id:
+        data['taskId'] = task_id
+    return ParseDict(data, Message())
+
+
+class GatewayExecutor(AgentExecutor):
+    """Wrap any native async handler as an A2A task, without detached worker hacks."""
+
+    def __init__(self, handler, task_mode=True, timeout_seconds=120.0):
+        self.handler = handler
+        self.task_mode = task_mode
+        self.timeout_seconds = timeout_seconds
+
+    async def execute(self, context, event_queue):
+        message = context.message
+        if message is None or message.role != Role.ROLE_USER:
+            raise InvalidParamsError('Expected a user message')
+        payload = {}
+        for part in MessageToDict(message).get('parts', []):
+            if 'data' in part:
+                if not isinstance(part['data'], dict):
+                    raise InvalidParamsError('Structured data must be an object')
+                payload.update(part['data'])
+            elif 'text' not in part:
+                raise UnsupportedOperationError('Only text and structured data are supported')
+        metadata = MessageToDict(message.metadata)
+        session_id = context.context_id or str(uuid4())
+        payload.update(message=context.get_user_input(), session_id=session_id)
+        request = {
+            'channel': 'web', 'payload': payload, 'session_id': session_id,
+            'tenant_id': metadata.get('tenant_id'), 'agent_id': metadata.get('agent_id'),
+            'metadata': metadata, 'backend_id': metadata.get('backend_id'),
+        }
+        if not self.task_mode:
+            async with asyncio.timeout(self.timeout_seconds):
+                response = await self.handler(request)
+            await event_queue.enqueue_event(_response_message(response, session_id))
+            return
+
+        updater = TaskUpdater(event_queue, context.task_id, session_id)
+        if context.current_task is None:
+            await event_queue.enqueue_event(Task(
+                id=context.task_id, context_id=session_id,
+                status=TaskStatus(state=TaskState.TASK_STATE_SUBMITTED),
+                history=[message],
+            ))
+        await updater.update_status(TaskState.TASK_STATE_WORKING)
+
+        async def progress(event, data):
+            await updater.update_status(TaskState.TASK_STATE_WORKING, metadata={
+                'framework_event': event, 'framework_payload': data,
+            })
+
+        token = _progress_callback.set(progress)
+        try:
+            async with asyncio.timeout(self.timeout_seconds):
+                response = await self.handler(request)
+            reply = _response_message(response, session_id, context.task_id)
+            response_metadata = response.get('metadata') or {}
+            requested_state = response_metadata.get('a2a_state') or response_metadata.get('a2a_task_state')
+            state = {
+                'input-required': TaskState.TASK_STATE_INPUT_REQUIRED,
+                'auth-required': TaskState.TASK_STATE_AUTH_REQUIRED,
+                'TASK_STATE_INPUT_REQUIRED': TaskState.TASK_STATE_INPUT_REQUIRED,
+                'TASK_STATE_AUTH_REQUIRED': TaskState.TASK_STATE_AUTH_REQUIRED,
+            }.get(requested_state, TaskState.TASK_STATE_COMPLETED)
+            await updater.add_artifact(list(reply.parts), artifact_id='response', name='Framework response', last_chunk=True)
+            await updater.update_status(state, message=reply)
+        except asyncio.CancelledError:
+            raise
+        except Exception as exc:
+            logger.exception('A2A task %s failed', context.task_id)
+            text = 'Processing timed out' if isinstance(exc, TimeoutError) else 'Agent processing failed'
+            await updater.update_status(TaskState.TASK_STATE_FAILED, message=_response_message({'text': text}, session_id, context.task_id))
+        finally:
+            _progress_callback.reset(token)
+
+    async def cancel(self, context, event_queue):
+        if not self.task_mode:
+            raise TaskNotCancelableError()
+        # The SDK cancels the running execute coroutine after this callback.
+        await TaskUpdater(event_queue, context.task_id, context.context_id).update_status(TaskState.TASK_STATE_CANCELED)
+
+
+def install_a2a(app, handler, registry, public_url, security_schemes=None, security=None,
+                *, task_mode=True, timeout_seconds=120.0, task_store=None,
+                name='Agent Framework OCI Gateway', skills=None):
+    """Reuse parent middleware. A persistent SDK TaskStore may be injected."""
+    if timeout_seconds <= 0:
+        raise ValueError('A2A timeout_seconds must be positive')
+    if skills is None:
+        skills = [{'id': b.backend_id, 'name': b.name or b.backend_id,
+                   'description': b.description or b.backend_id,
+                   'tags': b.domains or [b.backend_id], 'examples': b.examples}
+                  for b in registry.list()]
+    card = ParseDict({
+        'name': name, 'version': '1.1.0',
+        'description': 'Native framework agents and A2A interoperability',
+        'supportedInterfaces': [{'url': public_url.rstrip('/') + '/a2a',
+                                 'protocolBinding': 'JSONRPC', 'protocolVersion': '1.0'}],
+        'capabilities': {'streaming': task_mode, 'pushNotifications': False},
+        'defaultInputModes': ['text/plain', 'application/json'],
+        'defaultOutputModes': ['text/plain', 'application/json'],
+        'securitySchemes': security_schemes or {},
+        'securityRequirements': [{'schemes': {k: {'list': v} for k, v in item.items()}}
+                                 for item in (security or [])],
+        'skills': skills,
+    }, AgentCard())
+    request_handler = DefaultRequestHandler(
+        GatewayExecutor(handler, task_mode=task_mode, timeout_seconds=timeout_seconds),
+        task_store if task_store is not None else InMemoryTaskStore(), card,
+    )
+    add_a2a_routes_to_fastapi(
+        app, agent_card_routes=create_agent_card_routes(card),
+        jsonrpc_routes=create_jsonrpc_routes(
+            request_handler, rpc_url='/a2a', enable_v0_3_compat=True,
+            context_builder=FrameworkCallContextBuilder(),
+        ),
+    )
+    return request_handler
+
+
+def install_backend_a2a(app, handler, profiles, settings, *, task_store=None):
+    """Expose a framework backend directly, with skills built from agent profiles."""
+    return install_a2a(
+        app, handler, None, settings.A2A_PUBLIC_URL,
+        settings.A2A_SECURITY_SCHEMES or None, settings.A2A_SECURITY or None,
+        task_mode=settings.A2A_TASK_MODE, timeout_seconds=settings.A2A_TASK_TIMEOUT_SECONDS,
+        task_store=task_store, name=settings.APP_NAME,
+        skills=[{'id': p.agent_id, 'name': p.name or p.agent_id,
+                 'description': p.description or p.agent_id, 'tags': [p.agent_id]}
+                for p in profiles.list_profiles()],
+    )

libs/agent_framework/src/agent_framework/channels/interruption.py

@@ -1,5 +1,7 @@
 from __future__ import annotations
 
+from agent_framework.decisions import invoke as decision_invoke, predicate
+
 from dataclasses import dataclass
 from typing import Any
 
@@ -55,8 +57,11 @@ async def classify_processing_interruption(
         f"Complemento/interrupção atual: {supplement_client}\n"
     )
     try:
-        response = await llm.ainvoke(
+        response = await decision_invoke(llm,
             [{"role": "system", "content": prompt}],
+            decision_point="processing_interruption",
+            decision_questions=[predicate("regenerate", "Does the interruption require regenerating the response?")],
+            decision_render=lambda d: "1" if d["regenerate"] else "0",
             temperature=0,
             fallback_max_tokens=8,
             profile_name=profile_name,

libs/agent_framework/src/agent_framework/config/settings.py

@@ -14,6 +14,16 @@ load_dotenv(override=False)
 class Settings(BaseSettings):
     model_config = SettingsConfigDict(env_file='.env', env_file_encoding='utf-8', extra='ignore')
 
+    ENABLE_A2A: bool = False
+    A2A_PUBLIC_URL: str = 'http://localhost:8000'
+    A2A_TASK_MODE: bool = True
+    A2A_TASK_TIMEOUT_SECONDS: float = Field(default=120.0, gt=0)
+    A2A_SECURITY_SCHEMES: dict = Field(default_factory=dict)
+    A2A_SECURITY: list[dict[str, list[str]]] = Field(default_factory=list)
+
+    AGENT_LOCALE: str = 'pt-BR'
+    AGENT_LOCALES_DIR: str | None = None
+
     APP_NAME: str = 'ai-agent-template' 
     APP_ENV: str = 'local'
     LOG_LEVEL: str = 'INFO'
@@ -275,3 +285,7 @@ def get_settings() -> Settings:
     return Settings()
 
 settings = get_settings()
+
+# Fail early for invalid locale catalogs rather than during a transaction.
+from agent_framework.localization import catalog
+catalog(settings.AGENT_LOCALE)

libs/agent_framework/src/agent_framework/decisions/__init__.py

@@ -0,0 +1,2 @@
+"""Opt-in typed decision providers. Generative tasks keep their existing provider."""
+from .client import choice, predicate, score, invoke, selected, evaluate

libs/agent_framework/src/agent_framework/decisions/adapters/__init__.py

@@ -0,0 +1 @@
+"""Optional decision-provider adapters; no changes to the default engine."""

libs/agent_framework/src/agent_framework/decisions/adapters/typesafe.py

@@ -0,0 +1,97 @@
+"""Isolated TypeSafe API adapter. Uses the verified `noul` wire primitive.
+
+Choice is implemented as independent binary evaluations with strict rejection of
+ambiguous outcomes. This is an adapter strategy, NOT a native TypeSafe choice
+contract. The framework's original decision engine is never changed.
+"""
+from __future__ import annotations
+import os
+import logging
+import json
+
+logger = logging.getLogger("agent_framework.decisions")
+
+class TypeSafeContractError(ValueError):
+    def __init__(self, code: str):
+        self.code = code
+        super().__init__(code)
+
+
+def _wire_name(index: int, option: int | None = None) -> str:
+    return f"__jev_q{index}" if option is None else f"__jev_q{index}_o{option}"
+
+
+def build_request(state: str, questions: list[dict]) -> dict:
+    converted = {}
+    for i, question in enumerate(questions):
+        kind = question["type"]
+        if kind == "predicate":
+            converted[_wire_name(i)] = {"type": "noul", "instructions": question["instructions"]}
+        elif kind == "choice":
+            options = question["choices"]
+            if not options:
+                raise TypeSafeContractError("EMPTY_CHOICE")
+            for j, option in enumerate(options):
+                # Do not interpolate the user's state into the instruction.
+                converted[_wire_name(i, j)] = {
+                    "type": "noul",
+                    "instructions": (
+                        f"Assess whether '{option['value']}' is the correct selection for "
+                        f"the decision described here: {question['instructions']}. "
+                        "Evaluate the supplied state and policy; answer yes only if this "
+                        "specific option is appropriate."
+                    ),
+                }
+        elif kind == "score":
+            # The direct API score contract has not been verified. Fail closed.
+            raise TypeSafeContractError("UNSUPPORTED_DIRECT_SCORE")
+        else:
+            raise TypeSafeContractError("UNSUPPORTED_QUESTION_TYPE")
+    return {"model": os.getenv("JEV_MODEL", "jev-latest"), "state": state, "questions": converted}
+
+
+def _probability(answer: object) -> float:
+    if not isinstance(answer, dict) or answer.get("type") not in (None, "noul", "predicate"):
+        raise TypeSafeContractError("INVALID_NOUL_ANSWER")
+    value = answer.get("noul", answer.get("probability"))
+    if isinstance(value, bool) or not isinstance(value, (int, float)) or not 0 <= value <= 1:
+        raise TypeSafeContractError("INVALID_NOUL_PROBABILITY")
+    return float(value)
+
+
+def normalize_answers(response: dict, questions: list[dict]) -> list[dict]:
+    if not isinstance(response, dict) or not isinstance(response.get("answers"), dict):
+        raise TypeSafeContractError("INVALID_ANSWERS_CONTAINER")
+    answers = response["answers"]
+    expected = set()
+    for i, q in enumerate(questions):
+        if q["type"] == "choice":
+            expected.update(_wire_name(i, j) for j in range(len(q["choices"])))
+        else:
+            expected.add(_wire_name(i))
+    if set(answers) != expected:
+        raise TypeSafeContractError("MISSING_OR_EXTRA_ANSWER")
+    result = []
+    for i, q in enumerate(questions):
+        kind = q["type"]
+        if kind == "predicate":
+            result.append({"name": q["name"], "type": "predicate", "probability": _probability(answers[_wire_name(i)])})
+        elif kind == "choice":
+            probabilities = [_probability(answers[_wire_name(i, j)]) for j in range(len(q["choices"]))]
+            ranked = sorted(range(len(probabilities)), key=lambda j: probabilities[j], reverse=True)
+            winner, runner_up = probabilities[ranked[0]], probabilities[ranked[1]] if len(ranked) > 1 else 0.0
+            # Require a single positive, unambiguous choice. Low confidence
+            # remains subject to the unchanged JEV_MIN_CONFIDENCE in client.py.
+            if os.getenv("JEV_DECISION_DIAGNOSTICS", "false").strip().lower() == "true":
+                logger.info("jev.choice.diagnostics question_index=%d probabilities=%s winner_index=%d winner=%.4f runner_up=%.4f margin=%.4f status=%s",
+                            i, json.dumps([round(v, 4) for v in probabilities]), ranked[0],
+                            winner, runner_up, winner - runner_up,
+                            "ambiguous" if winner <= 0.5 or winner - runner_up < 0.15 else "candidate")
+            if winner <= 0.5 or winner - runner_up < 0.15:
+                raise TypeSafeContractError("AMBIGUOUS_DIRECT_CHOICE")
+            result.append({"name": q["name"], "type": "choice",
+                           "choice": q["choices"][ranked[0]]["value"], "confidence": winner,
+                           "option_probabilities": probabilities})
+        else:
+            raise TypeSafeContractError("UNSUPPORTED_DIRECT_QUESTION_TYPE")
+    return result

libs/agent_framework/src/agent_framework/decisions/client.py

@@ -0,0 +1,298 @@
+from __future__ import annotations
+
+import asyncio
+import json
+import logging
+import math
+import os
+import re
+import time
+import urllib.request
+import urllib.error
+import urllib.parse
+from typing import Any
+
+logger = logging.getLogger("agent_framework.decisions")
+
+
+def env_key(point: str) -> str:
+    return "DECISION_" + re.sub(r"[^A-Z0-9]+", "_", point.upper()).strip("_") + "_PROVIDER"
+
+
+def selected(point: str, group: str | None = None) -> bool:
+    default = os.getenv(env_key(group), os.getenv("DECISION_PROVIDER", "current")) if group else os.getenv("DECISION_PROVIDER", "current")
+    value = os.getenv(env_key(point), default).strip().lower()
+    if value not in {"current", "jev"}:
+        raise ValueError(f"{env_key(point)} must be current or jev")
+    return value == "jev"
+
+
+def choice(name: str, values: list, instructions: str = "") -> dict:
+    values = list(dict.fromkeys(str(v) for v in values))
+    return {"type": "choice", "name": name, "instructions": instructions or f"Select {name} following the supplied policy.",
+            "choices": [{"value": v, "description": v} for v in values]}
+
+
+def predicate(name: str, instructions: str = "") -> dict:
+    return {"type": "predicate", "name": name, "instructions": instructions or f"Is {name} true under the supplied policy?"}
+
+
+def score(name: str, instructions: str = "") -> dict:
+    return {"type": "score", "name": name, "instructions": instructions or "Apply the supplied rubric on a 0 to 10 scale.",
+            "levels": [{"label": str(i)} for i in range(11)]}
+
+
+def _number(value: Any, low: float, high: float) -> float:
+    if isinstance(value, bool) or not isinstance(value, (int, float)) or not math.isfinite(value) or not low <= value <= high:
+        raise _validation_error("INVALID_PROBABILITY_OR_SCORE")
+    return float(value)
+
+
+def _jev_endpoint() -> str:
+    """Return a trusted HTTPS decision endpoint configured by the deployer.
+
+    The endpoint is an operator setting, never derived from model output.
+    Set JEV_ALLOWED_HOSTS in production to restrict credential destinations.
+    """
+    endpoint = os.getenv("JEV_ENDPOINT", "https://ai-gateway.vercel.sh/v1/decisions").strip()
+    parsed = urllib.parse.urlsplit(endpoint)
+    if (parsed.scheme != "https" or not parsed.hostname or parsed.username or
+            parsed.password or parsed.fragment or not parsed.path):
+        raise ValueError("JEV_ENDPOINT must be a valid HTTPS endpoint URL")
+    allowed = [host.strip().lower() for host in os.getenv(
+        "JEV_ALLOWED_HOSTS", "ai-gateway.vercel.sh,api.typesafe.ai").split(",") if host.strip()]
+    if parsed.hostname.lower() not in allowed:
+        raise ValueError("JEV_ENDPOINT host is not in JEV_ALLOWED_HOSTS")
+    return endpoint
+
+
+def _post(body: dict) -> dict:
+    key = (os.getenv("JEV_API_KEY") or os.getenv("AI_GATEWAY_API_KEY") or "").strip()
+    if not key:
+        raise ValueError("JEV_API_KEY or AI_GATEWAY_API_KEY is required for JEV")
+    timeout = float(os.getenv("JEV_TIMEOUT_SECONDS", "15"))
+    if not math.isfinite(timeout) or timeout <= 0:
+        raise ValueError("JEV_TIMEOUT_SECONDS must be positive")
+    request = urllib.request.Request(_jev_endpoint(),
+        data=json.dumps(body, ensure_ascii=False).encode("utf-8"), method="POST",
+        headers={"Authorization": "Bearer " + key, "Content-Type": "application/json"})
+    # urllib does not automatically follow redirects for POST in all cases;
+    # disable redirects to avoid sending credentials to another host.
+    class NoRedirect(urllib.request.HTTPRedirectHandler):
+        def redirect_request(self, req, fp, code, msg, headers, newurl):
+            return None
+
+    opener = urllib.request.build_opener(NoRedirect)
+    with opener.open(request, timeout=timeout) as response:
+        return json.load(response)
+
+
+def _safe_http_diagnostic(exc: urllib.error.HTTPError) -> str:
+    """Extract only structural validation hints; never log echoed inputs or tokens."""
+    if exc.code not in (400, 422):
+        return "unavailable"
+    try:
+        raw = exc.read(16384)
+        payload = json.loads(raw)
+    except (ValueError, TypeError, OSError, UnicodeError):
+        return "unavailable"
+    if not isinstance(payload, dict):
+        return "unavailable"
+    detail = payload.get("detail")
+    if not isinstance(detail, list):
+        return "unavailable"
+    safe = []
+    for item in detail[:5]:
+        if not isinstance(item, dict):
+            continue
+        # Only the Pydantic error *type* is safe to expose, not message/input/context.
+        kind = item.get("type")
+        if not isinstance(kind, str) or not re.fullmatch(r"[a-z][a-z0-9_]{0,63}", kind):
+            continue
+        loc = item.get("loc")
+        # Restrict locations to known schema tokens; never echo dynamic names.
+        known = {"body", "model", "state", "questions", "type", "instructions",
+                 "choices", "levels", "value", "description", "label"}
+        path = ".".join(part for part in loc if isinstance(part, str) and part in known) if isinstance(loc, list) else ""
+        safe.append(kind + (":" + path if path else ""))
+    return ",".join(safe) or "unavailable"
+
+
+def _diagnostics_enabled() -> bool:
+    """Opt-in numeric decision diagnostics; never expose state, prompts or choice text."""
+    return os.getenv("JEV_DECISION_DIAGNOSTICS", "false").strip().lower() == "true"
+
+
+def _validation_error(code: str) -> ValueError:
+    # Explicit, stable codes for logs; existing fallback/exception behavior stays unchanged.
+    return ValueError(code)
+
+
+async def evaluate(point: str, messages: list, questions: list[dict], *, llm=None, operation=None, group=None) -> dict:
+    if not questions or len({q["name"] for q in questions}) != len(questions):
+        raise ValueError("Decision questions must have unique names")
+    if any(q["type"] == "choice" and not q["choices"] for q in questions):
+        raise ValueError("Decision choice requires options")
+    started = time.monotonic()
+    state = json.dumps({"policy_and_context": messages}, ensure_ascii=False, default=str)
+    # The default Vercel wire contract remains byte-for-byte equivalent.
+    direct_typesafe = urllib.parse.urlsplit(_jev_endpoint()).hostname == "api.typesafe.ai"
+    if direct_typesafe:
+        from .adapters.typesafe import build_request, normalize_answers
+        request_body = build_request(state, questions)
+    else:
+        request_body = {"model": os.getenv("JEV_MODEL", "typesafe-ai/jev"),
+                        "input": state, "questions": questions}
+    response = await asyncio.to_thread(_post, request_body)
+    await _record(llm, point, response, operation)
+    answers = normalize_answers(response, questions) if direct_typesafe else response.get("answers")
+    if not isinstance(answers, list) or len(answers) != len(questions):
+        raise _validation_error("MISSING_DECISION_ANSWERS")
+    indexed = {a.get("name"): a for a in answers if isinstance(a, dict)}
+    if len(indexed) != len(questions):
+        raise _validation_error("DUPLICATE_OR_INVALID_ANSWERS")
+    result, confidences, predicate_probabilities = {}, [], []
+    diagnostic_questions = []
+    for q in questions:
+        a = indexed.get(q["name"], {})
+        if a.get("type") != q["type"]:
+            raise _validation_error("ANSWER_TYPE_MISMATCH")
+        if q["type"] == "choice":
+            value = a.get("choice")
+            if value not in {v["value"] for v in q["choices"]}:
+                raise _validation_error("UNDECLARED_CHOICE")
+            # Missing confidence must never be promoted to certainty.
+            confidences.append(_number(a.get("confidence", 0), 0, 1))
+            if _diagnostics_enabled():
+                options = [v["value"] for v in q["choices"]]
+                detail = {"question_index": len(diagnostic_questions), "type": "choice",
+                          "selected_option_index": options.index(value),
+                          "confidence": round(confidences[-1], 4)}
+                if isinstance(a.get("option_probabilities"), list):
+                    detail["option_probabilities"] = [round(_number(v, 0, 1), 4) for v in a["option_probabilities"]]
+                diagnostic_questions.append(detail)
+        elif q["type"] == "predicate":
+            probability = _number(a.get("probability"), 0, 1)
+            value = probability >= 0.5
+            confidences.append(max(probability, 1-probability))
+            predicate_probabilities.append((q["name"], probability))
+            if _diagnostics_enabled():
+                diagnostic_questions.append({"question_index": len(diagnostic_questions), "type": "predicate",
+                                             "decision": value, "probability_true": round(probability, 4),
+                                             "confidence": round(confidences[-1], 4)})
+        else:
+            value = _number(a.get("score"), 0, len(q["levels"])-1)
+            confidences.append(_number(a.get("confidence", 0), 0, 1))
+            if _diagnostics_enabled():
+                diagnostic_questions.append({"question_index": len(diagnostic_questions), "type": "score",
+                                             "score": value, "confidence": round(confidences[-1], 4)})
+        result[q["name"]] = value
+    confidence = min(confidences)
+    default_minimum = os.getenv(env_key(group).replace("_PROVIDER", "_MIN_CONFIDENCE"), os.getenv("JEV_MIN_CONFIDENCE", "0.7")) if group else os.getenv("JEV_MIN_CONFIDENCE", "0.7")
+    minimum = float(os.getenv(env_key(point).replace("_PROVIDER", "_MIN_CONFIDENCE"), default_minimum))
+    _number(minimum, 0, 1)
+    if _diagnostics_enabled():
+        logger.info("jev.decision.diagnostics point=%s status=%s confidence=%.4f minimum_confidence=%.4f questions=%s",
+                    point, "accepted" if confidence >= minimum else "rejected_low_confidence",
+                    confidence, minimum, json.dumps(diagnostic_questions, separators=(",", ":")))
+    if confidence < minimum:
+        # A NOUL value is P(True), not confidence in the True answer.
+        # Log only numeric calibration data, never decision input or question text.
+        logger.warning("decision point=%s provider=jev confidence=%.4f minimum_confidence=%.4f status=rejected predicate_probabilities=%s",
+                       point, confidence, minimum,
+                       [(name, round(probability, 4)) for name, probability in predicate_probabilities])
+        raise _validation_error("CONFIDENCE_BELOW_THRESHOLD")
+    result["confidence"] = confidence
+    result["reason"] = "JEV typed decision (no generated explanation)."
+    logger.info("decision point=%s provider=jev model=%s elapsed_ms=%.1f usage=%s gateway=%s",
+        point, response.get("model"), (time.monotonic()-started)*1000,
+        response.get("usage"), response.get("provider_metadata", {}).get("gateway", {}))
+    return result
+
+
+async def invoke(llm: Any, messages: list, *, decision_point: str, decision_questions: list,
+                 decision_render=None, decision_generative: bool = False, decision_group: str | None = None, **kwargs: Any) -> str:
+    if decision_generative or not selected(decision_point, decision_group):
+        return await llm.ainvoke(messages, **kwargs)
+    try:
+        result = await evaluate(decision_point, messages, decision_questions, llm=llm, operation=kwargs.get("generation_name"), group=decision_group)
+        return decision_render(result) if decision_render else json.dumps(result, ensure_ascii=False)
+    except Exception as exc:
+        # Do not log exception text, input, response bodies or credentials.
+        from urllib.error import HTTPError
+        from .adapters.typesafe import TypeSafeContractError
+        category = (exc.code if isinstance(exc, TypeSafeContractError) else
+                    "HTTP_ERROR" if isinstance(exc, HTTPError) else
+                    "VALIDATION_ERROR" if isinstance(exc, ValueError) else
+                    "TRANSPORT_ERROR")
+        http_status = exc.code if isinstance(exc, HTTPError) else None
+        error_code = (exc.code if isinstance(exc, TypeSafeContractError) else
+                      str(exc) if isinstance(exc, ValueError) and
+                      str(exc) in {"MISSING_DECISION_ANSWERS", "DUPLICATE_OR_INVALID_ANSWERS",
+                                   "ANSWER_TYPE_MISMATCH", "UNDECLARED_CHOICE",
+                                   "CONFIDENCE_BELOW_THRESHOLD", "INVALID_PROBABILITY_OR_SCORE"}
+                      else "UNCLASSIFIED")
+        http_detail = _safe_http_diagnostic(exc) if isinstance(exc, HTTPError) else "not_applicable"
+        logger.warning("decision point=%s provider=jev failure=%s category=%s http_status=%s "
+                       "error_code=%s http_schema=%s question_types=%s",
+                       decision_point, type(exc).__name__, category, http_status,
+                       error_code, http_detail, ",".join(sorted({q.get("type", "unknown") for q in decision_questions if isinstance(q, dict)})))
+        fallback = os.getenv("DECISION_FALLBACK_TO_CURRENT", "true").strip().lower()
+        if fallback not in {"true", "false"}:
+            raise ValueError("DECISION_FALLBACK_TO_CURRENT must be true or false")
+        if fallback != "true":
+            raise
+        logger.info("decision point=%s provider=current fallback=true", decision_point)
+        return await llm.ainvoke(messages, **kwargs)
+
+
+async def multi_intent(llm, messages, *, intents, text, **kwargs):
+    """Keep source-fragment extraction and request order; JEV selects each intent."""
+    raw = await llm.ainvoke(messages, **kwargs)
+    if not selected("multi_intent"):
+        return raw
+    from agent_framework.llm.structured_output import parse_json_object
+    candidates = parse_json_object(raw).get("operations")
+    if not isinstance(candidates, list) or not candidates:
+        raise ValueError("Multi-intent extraction returned no candidates")
+    if any(not isinstance(c, dict) or not str(c.get("source_text") or "").strip()
+           or str(c["source_text"]) not in text for c in candidates):
+        raise ValueError("Multi-intent source must be a literal user-message fragment")
+    questions = [choice("operation_" + str(n), [i.name for i in intents] + ["__unsupported__"],
+        "Classify ONLY this extracted objective: " + c["source_text"] + ". Use __unsupported__ for an objective outside the intent catalog.") for n, c in enumerate(candidates)]
+
+    class ExtractedProvider:
+        usage_repository = getattr(llm, "usage_repository", None)
+        telemetry = getattr(llm, "telemetry", None)
+
+        async def ainvoke(self, messages, **kwargs):
+            return raw
+
+    return await invoke(ExtractedProvider(), messages, decision_point="multi_intent", decision_questions=questions,
+        decision_render=lambda d: json.dumps({"confidence": d["confidence"], "operations": [
+            {"intent": None if d["operation_" + str(n)] == "__unsupported__" else d["operation_" + str(n)],
+             "unsupported": d["operation_" + str(n)] == "__unsupported__", "source_text": c["source_text"]}
+            for n, c in enumerate(candidates)]}))
+
+
+async def _record(llm, point, response, operation):
+    """Account for billed responses even if validation later rejects them."""
+    if llm is None:
+        return
+    usage = response.get("usage") or {}
+    gateway = (response.get("provider_metadata") or {}).get("gateway") or {}
+    model = response.get("model") or os.getenv("JEV_MODEL", "typesafe-ai/jev")
+    metadata = {"provider": "jev", "model": model, "decision_point": point,
+                "generation_id": gateway.get("generationId"), "gateway_cost": gateway.get("cost")}
+    try:
+        repository = getattr(llm, "usage_repository", None)
+        if repository is not None:
+            from agent_framework.billing.usage_repository import UsageRecord
+            await repository.record(UsageRecord.from_usage("jev", model, operation or "decision." + point,
+                {"prompt_tokens": usage.get("input_tokens", 0), "completion_tokens": usage.get("output_tokens", 0),
+                 "total_tokens": usage.get("total_tokens", 0), "cost_usd": gateway.get("cost", 0)}, metadata))
+        telemetry = getattr(llm, "telemetry", None)
+        if telemetry is not None:
+            await telemetry.event("decision.jev", {**metadata, "usage": usage})
+    except Exception as exc:
+        logger.warning("decision accounting point=%s failure=%s", point, type(exc).__name__)

libs/agent_framework/src/agent_framework/global_supervisor/client.py

@@ -1,6 +1,8 @@
 from __future__ import annotations
 
 import time
+import os
+from collections import OrderedDict
 from typing import Any
 
 import httpx
@@ -11,6 +13,8 @@ from .models import BackendCallResult, BackendDefinition, GlobalRouteDecision
 class BackendClient:
     def __init__(self, timeout_seconds: float = 120.0):
         self.timeout_seconds = timeout_seconds
+        # Bounded, process-local continuation cache; native calls do not use it.
+        self._a2a_sessions = OrderedDict()
 
     async def call_message(
         self,
@@ -19,6 +23,27 @@ class BackendClient:
         route_decision: GlobalRouteDecision,
         use_sse: bool = False,
     ) -> BackendCallResult:
+        if backend.protocol == "a2a":
+            from agent_framework.a2a.client import call_a2a
+            started = time.monotonic()
+            payload = dict(request_payload.get("payload") or {})
+            key = (request_payload.get("tenant_id") or "default", backend.backend_id, payload.get("session_id"))
+            previous = self._a2a_sessions.get(key) if key[2] else None
+            if previous and time.monotonic() - previous[0] < 3600:
+                payload.setdefault("a2a_context_id", previous[1].get("a2a_context_id"))
+                if previous[1].get("a2a_state") in {"input-required", "auth-required", "TASK_STATE_INPUT_REQUIRED", "TASK_STATE_AUTH_REQUIRED"}:
+                    payload.setdefault("a2a_task_id", previous[1].get("a2a_task_id"))
+            data = await call_a2a(backend, {**request_payload, "payload": payload}, self.timeout_seconds)
+            if key[2]:
+                self._a2a_sessions[key] = (time.monotonic(), data.get("metadata", {}))
+                self._a2a_sessions.move_to_end(key)
+                while len(self._a2a_sessions) > 10000:
+                    self._a2a_sessions.popitem(last=False)
+            return BackendCallResult(
+                backend_id=backend.backend_id, backend_url=backend.base_url,
+                status_code=200, response=data, route_decision=route_decision,
+                elapsed_ms=int((time.monotonic() - started) * 1000),
+            )
         path = backend.sse_message_path if use_sse else backend.message_path
         url = f"{backend.base_url}{path}"
         payload = dict(request_payload)
@@ -45,8 +70,16 @@ class BackendClient:
         )
 
     async def health(self, backend: BackendDefinition) -> dict[str, Any]:
-        url = f"{backend.base_url}{backend.health_path}"
-        async with httpx.AsyncClient(timeout=10.0) as client:
+        path = backend.a2a_agent_card_path if backend.protocol == "a2a" else backend.health_path
+        url = f"{backend.base_url}{path}"
+        headers = {}
+        if backend.protocol == "a2a":
+            for header, env_name in backend.a2a_headers_env.items():
+                value = os.environ.get(env_name)
+                if not value:
+                    return {"backend_id": backend.backend_id, "ok": False, "error": f"Missing credential environment variable: {env_name}"}
+                headers[header] = value
+        async with httpx.AsyncClient(timeout=10.0, headers=headers) as client:
             try:
                 resp = await client.get(url)
                 return {"backend_id": backend.backend_id, "status_code": resp.status_code, "ok": resp.is_success, "body": self._safe_json(resp)}

libs/agent_framework/src/agent_framework/global_supervisor/models.py

@@ -21,6 +21,12 @@ class BackendDefinition(BaseModel):
     examples: list[str] = Field(default_factory=list)
     priority: int = 100
     enabled: bool = True
+    protocol: Literal["native", "a2a"] = "native"
+    a2a_version: Literal["1.0", "0.3"] = "1.0"
+    a2a_agent_card_path: str = "/.well-known/agent-card.json"
+    a2a_rpc_path: str = "/a2a"
+    a2a_headers_env: dict[str, str] = Field(default_factory=dict)
+    a2a_poll_interval_seconds: float = Field(default=0.5, gt=0)
     health_path: str = "/health"
     message_path: str = "/gateway/message"
     sse_message_path: str = "/gateway/message/sse"

libs/agent_framework/src/agent_framework/global_supervisor/router.py

@@ -1,5 +1,9 @@
 from __future__ import annotations
 
+from agent_framework.decisions import invoke as decision_invoke, choice
+
+from agent_framework.localization import terms
+
 from agent_framework.llm.structured_output import parse_json_object
 
 import json
@@ -13,9 +17,7 @@ from .session_store import InMemoryGlobalSessionStore
 
 logger = logging.getLogger("agent_framework.global_supervisor")
 
-_TERMINAL_WORDS = {
-    "obrigado", "obrigada", "valeu", "tchau", "encerrar", "fim", "cancelar atendimento"
-}
+# Terminal vocabulary is read from the selected locale at request time.
 
 
 class GlobalSupervisorRouter:
@@ -162,10 +164,12 @@ class GlobalSupervisorRouter:
             return fallback or self._route_by_rules(text, mode)
         prompt = self._build_supervisor_prompt(text, request)
         try:
-            raw = await self.llm.ainvoke([
+            raw = await decision_invoke(self.llm, [
                 {"role": "system", "content": "Você é um supervisor global de backends. Responda somente JSON válido."},
                 {"role": "user", "content": prompt},
-            ], temperature=0, profile_name="supervisor", component_name="supervisor", generation_name="llm.supervisor")
+            ], temperature=0, decision_point='global_supervisor',
+                decision_questions=[choice("backend_id", list(self.registry.backends))],
+                profile_name="supervisor", component_name="supervisor", generation_name="llm.supervisor")
             data = self._parse_json(raw)
             backend_id = str(data.get("backend") or data.get("backend_id") or "").strip()
             if backend_id not in self.registry.backends:
@@ -213,10 +217,10 @@ class GlobalSupervisorRouter:
         n = self._normalize(text)
         if not n:
             return True
-        if n in _TERMINAL_WORDS:
+        if n in {self._normalize(value) for value in terms("global_router.terminal")}:
             return False
         tokens = n.split()
-        followup_markers = ["esse", "essa", "isso", "valor", "ele", "ela", "tambem", "e ", "entao", "nesse", "nessa"]
+        followup_markers = [self._normalize(value) for value in terms("global_router.followup")]
         return len(tokens) <= 6 or any(marker in n for marker in followup_markers)
 
     def _extract_text(self, request: GlobalRouteRequest) -> str: