Record denied API access without turning audit logs into a token store
Security needs failed-access context, but the existing logger captures the Authorization header and request body.
- Focused work estimate
- 2h 30m + prerequisites
- Priority in the scenario
- Medium
- Engineering practice
- Audit logging · Privacy
Estimated field mix
- Security50%
- Privacy engineering30%
- Site reliability20%
Field percentages are editorial estimates of the ticket's engineering focus. They total 100%; they are not measured time, proficiency scores, or ownership evidence.
Review it, then add it to your workspace.
The board opens an editable draft; nothing is saved until you confirm it. Sign-in and workspace permissions apply, and Demo boards remain ephemeral.
Project context
A fictional B2B export API uses long-lived tokens. Tokens copied between organizations can access too much, and revocation takes effect unpredictably.
Setup prerequisites
- Create a local synthetic API and two isolated organizations.
- Use generated disposable credentials only.
Preceding work
Complete these dependencies, or supply their agreed outputs before taking this ticket.
- ATOKEN-101 · Define token scopes as an explicit allowlist of export operations
- ATOKEN-102 · Store only token verification material and a safe display prefix
- ATOKEN-103 · Bind API token queries to the issuing organization
- ATOKEN-104 · Enforce token expiration at the request authorization boundary
- ATOKEN-105 · Rotate a token with a bounded overlap window
- ATOKEN-106 · Make revocation override cached authentication results
Acceptance criteria
- Record safe token identity, route, reason and UTC time.
- Exclude authorization headers and body content.
- Bound repeated denial events to prevent log exhaustion.
Implementation constraints
- Use synthetic token strings in redaction tests.
Verification to include
- Trace a scope denial by safe token identity.
- Send repeated malformed tokens and verify redaction and bounded logging.
Deliverables
- Safe denial audit and rate-bound checks
Rollout and recovery
Deploy audit filtering first; quarantine old diagnostic logs for authorized review.
Value of the work
For the engineer: Practice authorization boundaries, credential lifecycle and safe diagnostics.
For the team: Review denial paths and operational control over service access.
Evidence boundaries
Outcome Evidence: Tests, patches, and runbooks are requested deliverables. They become Outcome Evidence only through a qualified Mission and immutable Evidence IDs.
Ownership Evidence: Independent adaptation must be observed under a declared verification policy and cite immutable Evidence IDs. Completing a planning ticket establishes no Ownership Evidence.