Skip to content

v0.13.0: MCP tools for Nova integration

Latest
Latest

Choose a tag to compare

View all tags
@sydney-nebulock sydney-nebulock released this 21 May 23:12
· 11 commits to main since this release
v0.13.0
9600f7d

Highlights

New MCP tools that unlock Nova's agentic research and investigation flows, plus a dryRun correctness fix in the automation loop.

New Features

  • athf_research_new MCP tool — deep research wrapper around HuntResearcherAgent + ResearchManager (Tavily + 5-skill methodology). Returns a structured JSON report consumable by Nova's intel triage. (4fc68af)
  • athf_investigate_new MCP tool — investigation creation entrypoint for Nova chat integration. (5c9b240)

Fixes

  • automation/loop.ts — dryRun no longer mutates state. Previously state.issues / state.prs writes and saveState() calls fired regardless of dryRun; now every state mutation is gated on !dryRun. (8079d91)
  • attack_tools module restored — server.py referenced a missing module. (418ebb4)

Docs

  • LLM provider configuration guide — getting-started.md now documents Anthropic, OpenAI, Bedrock, Ollama, and LiteLLM auto-detection plus the explicit ATHF_LLM_PROVIDER / ATHF_LLM_MODEL override path. (5a19cd0)

Tests

  • Removed low-signal tests, strengthened provider contract assertions. (3499562)

Install

pip install agentic-threat-hunting-framework==0.13.0
Assets 2
Loading