Governance Quick Reference

Quick reference for the Mind Palace governance system.

Authority States

MCP Modes

CLI Commands

Proposals

# List proposals
palace proposals
palace proposals --status proposed
palace proposals --status approved
palace proposals --status rejected
 
# Review
palace approve <id>
palace reject <id>
 
# Create
palace propose --type decision --content "Use JWT"

Knowledge

# With approval workflow
palace store --type decision --content "..." --scope palace
 
# Direct (approved immediately)
palace store-direct --type decision --content "..." --scope palace

MCP Tools

Agent Mode

• recall [--id ID] [--query TEXT] - Fetch approved learnings
• recall_decisions [--id ID] [--query TEXT] - Fetch approved decisions
• recall_learnings [--id ID] [--query TEXT] - Fetch approved learnings
• recall_fragments [--id ID] [--query TEXT] - Fetch approved fragments
• get_route --intent TEXT --scope SCOPE - Get navigation route
• store --type TYPE --content TEXT --scope SCOPE - Create proposal

Human Mode

All agent tools plus:

• store_direct --type TYPE --content TEXT --scope SCOPE - Create approved record
• approve_proposal --id ID - Approve proposal
• reject_proposal --id ID - Reject proposal

Scope Hierarchy

palace (global)
  └── project
      └── room (file/module)

Queries inherit from parent scopes.

Workflow Examples

Agent Discovery

{
  "tool": "get_route",
  "arguments": {
    "intent": "understand auth",
    "scope": "palace"
  }
}

Returns nodes with fetch_ref:

{
  "nodes": [
    {
      "id": "dec_123",
      "fetch_ref": "recall_decisions --id dec_123",
      "summary": "Use JWT"
    }
  ]
}

Create Proposal

{
  "tool": "store",
  "arguments": {
    "type": "decision",
    "content": "Use PostgreSQL",
    "scope": "palace"
  }
}

Approve Proposal

palace proposals --status proposed
palace approve prop_abc123

Direct Store (Human)

{
  "tool": "store_direct",
  "arguments": {
    "type": "decision",
    "content": "Use Redis for caching",
    "scope": "palace"
  }
}

Best Practices

For AI Agents:

• ✅ Use get_route for discovery
• ✅ Follow fetch_ref for details
• ✅ Create proposals via store
• ✅ Trust approved knowledge

For Humans:

• ✅ Review proposals regularly
• ✅ Approve quality content
• ✅ Reject incorrect/low-quality
• ✅ Use store_direct for verified info

Troubleshooting

Further Reading

• Full Governance Guide
• MCP Integration
• Implementation Logs