Locally-deployable AI agent for Perforce Helix Core (p4d) server monitoring,
root cause analysis, and governed corrective action.
Runs entirely on-premise. No cloud API calls during operation.
See docs/architecture.md for the layered design (signal extraction →
anomaly detection → SLM reasoning → action execution).
| Phase | Status | Description |
|---|---|---|
| 1 | In progress | Signal extraction & detection (no SLM) |
| 2 | Planned | SLM integration (Ollama / Phi-4) |
| 3 | Planned | RAG knowledge base |
| 4 | Planned | Annotation & eval harness |
pip install -r requirements.txt
pytest tests/Default action tier is Tier 1 (alert only). The agent never modifies
depot data and never acts without a written audit trail.
See docs/safety-policy.md and CLAUDE.md.
# p4-rca-agent Locally-deployable AI agent for Perforce Helix Core (`p4d`) server monitoring, root cause analysis, and governed corrective action. Runs entirely on-premise. No cloud API calls during operation. ## Design See `docs/architecture.md` for the layered design (signal extraction → anomaly detection → SLM reasoning → action execution). ## Implementation Phases | Phase | Status | Description | |---|---|---| | 1 | In progress | Signal extraction & detection (no SLM) | | 2 | Planned | SLM integration (Ollama / Phi-4) | | 3 | Planned | RAG knowledge base | | 4 | Planned | Annotation & eval harness | ## Quick Start ```bash pip install -r requirements.txt pytest tests/ ``` ## Safety Default action tier is **Tier 1 (alert only)**. The agent never modifies depot data and never acts without a written audit trail. See `docs/safety-policy.md` and `CLAUDE.md`.
| # | Change | User | Description | Committed | |
|---|---|---|---|---|---|
| #1 | 32636 | bot_Claude_Anthropic |
Scaffold p4-rca-agent repo: directory structure, data models, layer stubs, test fixtures, config, docs. Covers briefing tasks 2 and 3. #review-32637 @robert_cowham @tom_tyler |