SilentAuth is the authorization control plane for autonomous systems. AI agents, CI/CD pipelines, and internal automation ask for permission before sensitive work, then verify a signed receipt before they act.

The core flow: Project → Agent → Intent → Policy → Receipt → Audit.

Quickstart

Get the full approval loop working before adding real connector credentials.

Create a project

Dashboard -> Projects -> New project

Start from the dashboard and create the scope that owns API keys, policies, agents, and audit records.

Register an agent

agent: deployment-runner
environment: production

Give every automation a stable identity before it can submit approval requests.

Submit an intent

POST https://api.silentauth.ai/api/intents
{ action, source_type, context }

The agent declares the exact action and parameters instead of executing first.

Verify the receipt

receipt_version: 2
params_hash: sha256:...
proof_tier: stub_vinacfm_l3

Execution happens only after the agent verifies a short-lived receipt for that exact request.

Intent Declaration →

How agents declare what they want to do

Policy Configuration →

Rules that determine when humans approve

Receipt Verification →

Verifying Receipt V2 claims before execution

AI Agent Integration →

Gate autonomous agent actions with human approval