Audit logging
Instrumented Steward routes emit queryable and exportable audit events for signing, proxy, policy, credential, and approval activity. Coverage is route-specific. The tenant-scoped HMAC chain and Ed25519 checkpoints make exported evidence tamper-evident after the bundled public key is separately matched to a trusted signing key or fingerprint.What’s Logged
Log Entry Structure
Every audit log entry includes:Transaction History
Every agent has a complete transaction history:- Status —
signed,pending,rejected - Policy results — full evaluation details
- TX hash — for broadcast transactions
- Timestamps — created, signed, resolved
Webhooks
Steward dispatches signed webhooks for key events. Create an endpoint through the webhook API and store the one-time secret returned in the response:approval_required
approval_required
tx_signed
tx_signed
tx_rejected
tx_rejected
tx_failed
tx_failed
Compliance
Steward’s audit log supports compliance requirements:- Tamper evidence: HMAC-linked events and Ed25519-signed checkpoints expose modification after the bundled public key is separately matched to a trusted signing key or fingerprint
- Retention: configurable per tenant
- Export: API endpoints for event export and signed evidence bundles
- Trust limit: an operator controlling all relevant keys can fabricate a self-consistent history
- Optional third-party time bound: RFC 3161 anchoring can prove a checkpoint existed no later than a trusted TSA time, narrowing but not eliminating the pre-anchor rewrite window
Related
- Architecture — How audit logging fits into the system
- Tenants API — Configure webhook URLs
- Vault API — Transaction history endpoints