Developer documentation

Production readiness checklist

A practical go-live and ongoing-operations checklist for an RCN integration.

Access and ownership

Named business, technical, security and privacy owners are recorded.
The live institution is ACTIVE and membership roles follow least privilege.
Every workload has its own live API key, approved scopes, allowlisted IPs and expiry/rotation owner.
Secrets are in a production secret manager; no live secret exists in source, CI output, browser storage or developer notes.
Emergency revocation and administrator recovery procedures have been rehearsed.

Subject matching uses stable identifiers and a documented duplicate-review path.
External account IDs and source references map deterministically to the ledger.
Every required write persists an idempotency key before transmission.
Event/state mappings, currency precision, timestamps and correction flow have business-owner sign-off.
NR, confidence, reason codes, disputes and stale data display correctly and do not become automatic declines.
Affordability and final lending decisions remain independent of RCN output.

Timeouts, bounded retries and jitter are configured; unsafe writes are never retried without the original idempotency key.
Request IDs, status, latency and endpoint are logged without credentials or sensitive bodies.
Rate-limit remaining/reset values feed proactive throttling and alerts.
Webhook signatures, timestamps and event IDs are verified before durable acceptance.
Failed webhook delivery, reporting health, dispute backlog and API errors have alerts with accountable responders.
Backups, restoration, incident response and dependency failure exercises have passed in staging.

Complete the sandbox fixture and record expected/actual evidence.
Run security, privacy and operational reviews with named approvals.
Deploy configuration with a new live key and verify readiness endpoints.
Process one controlled end-to-end subject, account, event and query.
Reconcile the ledger, audit log, reporting health and displayed profile.
Increase traffic gradually while watching error, latency, throttling and queue metrics.
Schedule the first credential rotation, access review, dispute review and disaster-recovery exercise.

Developer documentation

A practical go-live and ongoing-operations checklist for an RCN integration.

Named business, technical, security and privacy owners are recorded.
The live institution is ACTIVE and membership roles follow least privilege.
Every workload has its own live API key, approved scopes, allowlisted IPs and expiry/rotation owner.
Secrets are in a production secret manager; no live secret exists in source, CI output, browser storage or developer notes.
Emergency revocation and administrator recovery procedures have been rehearsed.

Subject matching uses stable identifiers and a documented duplicate-review path.
External account IDs and source references map deterministically to the ledger.
Every required write persists an idempotency key before transmission.
Event/state mappings, currency precision, timestamps and correction flow have business-owner sign-off.
NR, confidence, reason codes, disputes and stale data display correctly and do not become automatic declines.
Affordability and final lending decisions remain independent of RCN output.

Timeouts, bounded retries and jitter are configured; unsafe writes are never retried without the original idempotency key.
Request IDs, status, latency and endpoint are logged without credentials or sensitive bodies.
Rate-limit remaining/reset values feed proactive throttling and alerts.
Webhook signatures, timestamps and event IDs are verified before durable acceptance.
Failed webhook delivery, reporting health, dispute backlog and API errors have alerts with accountable responders.
Backups, restoration, incident response and dependency failure exercises have passed in staging.

Complete the sandbox fixture and record expected/actual evidence.
Run security, privacy and operational reviews with named approvals.
Deploy configuration with a new live key and verify readiness endpoints.
Process one controlled end-to-end subject, account, event and query.
Reconcile the ledger, audit log, reporting health and displayed profile.
Increase traffic gradually while watching error, latency, throttling and queue metrics.
Schedule the first credential rotation, access review, dispute review and disaster-recovery exercise.