| name | open-brain |
| description | Axon MCP: search OpenOS docs and graph knowledge. |
| version | 2.1.0 |
| author | OpenPro |
| license | MIT |
| platforms | ["linux","macos","windows"] |
| metadata | {"hermes":{"tags":["openbrain","axon","mcp","rag","openos"],"category":"open-ecosystem","related_skills":["open-contract","open-ecosystem-hub","open-memory","open-brain-orchestrator","openbrain-chat-mermaid"]}} |
Open Brain (Axon)
Company Brain: graph, doc RAG, Skills File, MCP Knowledge.
When to Use
- "How does CC-W4-005 work?" — before guessing mesh behavior
- Cross-app feature — query indexed OpenOS docs first
- Validated institutional knowledge vs raw specs
For session-local recall only → open-memory first.
Prerequisites
mcp_servers:
openbrain-knowledge:
url: ${OPENBRAIN_API_URL}/api/v1/mcp/knowledge
headers:
Authorization: Bearer ${AXON_AGENT_API_KEY}
Env: OPENBRAIN_API_URL, AXON_AGENT_API_KEY — see OpenAgents/docs/openos-mesh-env.md.
Structural overview
| Surface | Use for |
|---|
search_knowledge + domain: openos | Specs, ADRs, registry, READMEs |
search_knowledge (no domain) | Validated graph nodes |
search_observations + app | Runtime what-happened |
get_skills_file | Agent skills export |
ingest_observation | Channel B summaries (non-blocking) |
Procedure
search_knowledge(query, domain="openos", app=optional, limit=8)
- Cite
path from hits in tickets/PRs
- For contracts →
open-contract handbook + registry paths from citations
- After doc edits →
brain-sync-docs.sh (maintainer)
Example:
{"query": "CC-W4-005 envelope", "domain": "openos", "app": "opencode", "limit": 8}
Decision rules
| Question type | Tool |
|---|
| Mesh spec / CC-* | search_knowledge domain openos |
| What happened in app | search_observations |
| Store new runtime fact | ingest_observation (no secrets/PII) |
Pitfalls
- Guessing contract IDs without search
- Storing secrets in observations
- Using Brain as RecEvent authority (use
open-rec for audit)
Verification