brain_claims
List active locks
How to use it
brain_claims is exposed by the Brain MCP MCP server. Add the server to your MCP client (Claude Desktop, Cursor, Windsurf and others), and the brain_claims tool becomes available to the model automatically. See the full listing for setup details and every tool this server provides.
Install Brain MCP
pip install -e .Other tools in Brain MCP (35)
Live health of all agents (status, heartbeat age, claims)
Run gate in a loop, wait for fixes, retry until clean
Save full working state
Recover after context compression
Lock a file/resource (TTL-based mutex)
Read the ledger
Log action/discovery/decision/error
Condensed view for context recovery
Validate all contracts — catches param mismatches, missing functions
Read other agents' contracts before coding
Publish what your module provides / expects
Direct message another agent
Remove outdated memories
Run compile + contract check, DM errors to responsible agents
Read your DMs
1.6ms
Success rates, duration, error counts per agent
Create a task DAG with dependencies
Get tasks whose dependencies are satisfied
Overall progress
Mark task done/failed (auto-promotes dependents)
Post to a channel
Heartbeat with status + progress (returns pending DMs)
Read from a channel
Search memories from previous sessions
Name this session
Unlock
Store persistent knowledge (survives brainclear)
Replace a failed agent with recovery context
List active sessions
Show session info + room
Spawn multiple agents in one call
Spawn a new agent (hermes, claude, or headless)
Persist the compiled workflow into brain state + a task DAG, optionally write conductor JSON
Turn one natural-language goal into phases, agents, file scopes, and conductor config