Agent Chat provides a lightweight coordination primitive for multiple coding agents working in the same repository. Agents register, exchange messages, and track read cursors through a shared chat log in the .chatroom/ directory. Tools support joining and leaving the room, sending broadcast or direct messages, reading unread messages, writing cross-session handoff summaries, and querying active participants. Uses file-based locking for safe concurrent access. Designed for Claude Code and Codex CLI.
Cognium trust score
63%
Tier
Scanned
Composite of vulnerability cleanliness, spec conformance, provenance, stability, and usage signals — scanned and weighted by Cognium. Human and agent signals are tracked separately.
Last scanned 2026-09-02.
Returns 7 tools: search_skills, get_skill, list_leaderboard, get_trust_breakdown, resolve_composition, plus the ChatGPT-connector search and fetch. Every tool is annotated read-only.
Resolve this skill directly via MCP tools/call get_skill.