Live action gates

Approval, policy, and sandbox boundaries for live sends, live spend, and high-risk operations.

Keyboard: / focus search Cmd/Ctrl+K open command menu
Status: implemented Version: latest Review: source-backed Last scanned: 2026-06-25T00:00:00Z Review required: false

Live action gates

Approval, policy, and sandbox boundaries for live sends, live spend, and high-risk operations.

Fail-closed defaults

Sensitive live actions fail closed unless account policy, RBAC, provider readiness, idempotency, content-firewall state, and durable approval records authorize execution.

Action Default Required gates
Live Gmail send/reply Blocked LIVE_OUTREACH, Gmail OAuth, sending alias, suppression check, safety review, approval
Agent spend/provisioning Blocked LIVE_SPEND, cap check, provider readiness, approval, idempotency
Runtime tool action Blocked if unsafe Project/account scope, content firewall, ToolRouter policy
Document/context release Quarantined Human review and source ref release
Skill promotion Draft/project-local Human approval and scope validation

Denied actions still produce auditable ToolCall and RunEvent proof without executing the adapter.

Official references

System Use in Trinity Official docs
Hermes Hosted agent runtime and skills context Hermes Agent docs
Jido Narrow policy/firewall action seam Jido Actions and Workflows
NVIDIA Nemotron scoring, safety, and QA decisions NIM LLM API reference
Stripe Checkout, webhooks, revenue proof, guarded spend Checkout Sessions API
Gmail Drafts, sends, aliases, scopes, inbound replies Gmail API scopes
Phoenix/Oban/Postgres Control plane, durable jobs, source-of-truth data Phoenix LiveView

Source paths

  • lib/autonomous_agency/policies
  • lib/autonomous_agency/tools/tool_router.ex
Was this page useful? Source-backed feedback keeps public docs honest.