GustavoNogueira/A-Team-Security-Infra-Agent-Engineer
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: async background chat, performance scaling, and dead code cleanup

Browse Source 
Async chat processing eliminates 504 timeouts - POST returns immediately,
backend processes in background, frontend polls for results with timestamps.
Scale to ~12 simultaneous chats via 8 uvicorn workers + 16-thread executor.
Parallelized MCP data collection, 2h session cache, 5min tool timeout.
Full dead code cleanup across backend, frontend, MCP, docker, and nginx.
This commit is contained in:
nogueiraguh
2026-03-06 11:06:21 -03:00
parent ef43eaa7ba
commit 2d024d3130
7 changed files with 272 additions and 265 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
README.md
View File

@@ -9,7 +9,7 @@
</p>
<p align="center">
<img src="https://img.shields.io/badge/version-1.9-C74634?style=flat-square" alt="Version">
<img src="https://img.shields.io/badge/version-2.0-C74634?style=flat-square" alt="Version">
<img src="https://img.shields.io/badge/python-3.12-3776AB?style=flat-square" alt="Python">
<img src="https://img.shields.io/badge/FastAPI-0.115-009688?style=flat-square" alt="FastAPI">
<img src="https://img.shields.io/badge/OCI-GenAI-C74634?style=flat-square" alt="OCI">
@@ -35,7 +35,8 @@ The platform combines security compliance scanning, AI-powered chat with **RAG (
- **MCP Tool Use (Function Calling)**: GenAI models can call tools from registered MCP servers during chat — supports both Cohere and Generic (OpenAI-style) function calling formats with automatic tool execution loop (max 5 iterations)
- **Chat Memory Compaction**: automatic summarization of older messages when conversation exceeds ~8000 tokens — keeps 6 recent messages intact and generates an LLM-based summary of older context, similar to Claude Code's context compression
- **Multimodal Chat**: upload images (PNG/JPG/GIF/WebP), PDFs, and text files directly in the chat for AI analysis — supports up to 5 files per message via OCI GenAI `ImageContent` and `DocumentContent`
- **Thinking Indicator**: button disables and shows spinner + "Pensando..." while waiting for GenAI response
- **Async Background Processing**: chat requests return immediately, GenAI + MCP tools process in background via dedicated thread pool (16 threads), frontend polls for results — eliminates 504 timeouts on long-running scans
- **Thinking Indicator**: button disables and shows spinner + "Pensando..." while waiting for GenAI response, with message timestamps
- 69 chat models + 11 embedding models across 6 providers: **Cohere**, **Meta**, **Google**, **OpenAI** (GPT-5.3/5.2/5.1/5/4.1/4o, Codex, Image, Audio, o1/o3/o4-mini, GPT-oss), **xAI** (Grok 4.1/4/3), **ProtectAI**
- OCID-based model resolution: catalog maps model IDs to OCI resource IDs per region for reliable API calls
- 16 OCI regions supported with auto-generated endpoints
@@ -76,7 +77,8 @@ The platform combines security compliance scanning, AI-powered chat with **RAG (
- `cis_get_scan_status` / `cis_invalidate_cache` — session status and cache management
- **Per-section data collection**: each scan tool collects only the OCI data needed for that section, avoiding unnecessary API calls
- **Region-specific scanning**: all scan tools accept optional `regions` parameter to target specific OCI regions (e.g., `["us-ashburn-1"]`) instead of scanning the entire tenancy
- **Session caching**: collected data is cached per config+regions scope, so subsequent scans on different sections reuse shared prerequisites (compartments, identity domains)
- **Session caching**: collected data is cached per config+regions scope (2-hour TTL), so subsequent scans on different sections reuse shared prerequisites (compartments, identity domains)
- **Parallelized data collection**: base collectors and regional collectors run in parallel thread pools (up to 8 workers), with 5-minute timeout per tool call
- Based on Oracle's official `cis_reports.py` (6660 lines, 48 CIS + 11 OBP checks)
### 🔌 MCP Server Registry + Tool Discovery
@@ -579,6 +581,7 @@ All models include OCID mapping for `us-ashburn-1`. For other regions, use the "
| Version | Date | Changes |
|---------|------|---------|
| **v2.0** | 2026-03 | Async background chat processing (no more 504 timeouts), frontend polling with timestamps, 8 uvicorn workers + 16-thread chat executor for ~12 simultaneous chats, parallelized MCP data collection (5-thread base + 8-thread regional), 2-hour MCP session cache, 5-min tool timeout, full dead code cleanup across backend/frontend/MCP |
| **v1.9** | 2026-03 | Multimodal chat (image/PDF/text file upload with OCI GenAI ImageContent/DocumentContent), region-specific MCP scanning (`regions` param on all scan tools), orphaned report auto-detection on progress poll, nginx timeout increased to 15min, improved API error handling for non-JSON responses |
| **v1.8** | 2026-03 | CIS Engine auto-update from Oracle GitHub with automatic patch reapplication, version check UI card (admin), new `/api/cis-engine/*` endpoints, report file listing and individual download endpoints, reorganized Reports tab (execution history + status) and Downloads tab (file browser only with expandable cards per report), CIS Level description tooltip, persistent log expand during report generation |
| **v1.7** | 2026-03 | Oracle official CIS report engine (replaces lightweight checker), granular report parameters (Level, OBP, Raw Data, Redact), per-report file storage with category browser, tenancy filter in Downloads, individual file download |