Headron Mind API

Declare the change class and any risk modifiers. The response names every gate that change owes and the policy revision behind them.

curl https://headron.tech/api/mind/plan \
  -H "Authorization: Bearer $MIND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"change_class": "api_change",
       "risk_modifiers": ["auth_boundary"]}'

{
  "change_class": "api_change",
  "risk_modifiers": ["auth_boundary"],
  "required_gates": [
    "formatter",
    "compiler_or_typechecker",
    "unit_tests",
    "schema_validation",
    "contract_tests",
    "endpoint_tests",
    "compatibility_check",
    "authorization_tests",
    "session_invalidation_test",
    "codeql",
    "secret_scan"
  ],
  "policy_revision": "310225fb7219aafe7ecf98c4e51d12d37249fc49",
  "scope": "Plan from caller declarations only; no project inspected and no gates run."
}

Seven change classes, from source_change to security_sensitive. Eleven risk modifiers, from auth_boundary to schema_migration. Each modifier adds the gates its risk owes.

A plan is not proof. Mind reads your declaration only: it does not inspect repositories or run checks. Your tools run the gates, and the engine keeps the receipt.

Point any MCP client at https://headron.tech/mcp. It serves public documentation to everyone and Mind planning to authenticated agents, inside the tool loop.

The open-source MCP server adds discovery for your own project.

$10 a month, billed monthly. Cancel anytime in your account; access runs through the paid period.

Generate a key in your account and copy the secret once. Keep it in your password manager or environment. Each request sends it in the Authorization header; Headron forwards it to Clerk to check entitlement and does not store it.

Key lost or leaked? Revoke it and generate a replacement.

The engine proves the change. Mind API tells the agent what to prove.