writing-skills
writing-skills
Use when creating new skills, editing existing skills, or verifying skills work before deployment
navan-deploy-integrationDeploy a Navan integration with tenant isolation, canary controls, and reversible data flow. Use when releasing an API, file, identity, or finance connector. Trigger with "deploy Navan integration", "release Navan connector", or "canary Navan sync".
Independent GitHub snapshot · Automated classification · Original documentation governs
Deploy a Navan integration with tenant isolation, canary controls, and reversible data flow. This workflow produces an auditable decision or artifact before any live action.
Deployment binds the application adapter to a documented Navan surface and an approved destination. Separate ingestion, normalization, publication, and write-back so each can stop without losing evidence.
Inject a production secret by reference at runtime; bind tenant, host, scope, destination, and environment. Verify revocation and break-glass procedures before traffic.
Use Read, Glob, and Grep to inspect documentation, schemas, configuration, code, fixtures, and evidence. Use Write and Edit only for approved repository artifacts. Invocation alone does not authorize network access, credentials, traveler or expense data, bookings, payments, policy or identity changes, file transfers, deployments, or deletion.
Production enablement, egress, data movement, scheduling, scope expansion, and write-back require explicit service and data-owner approval.
Return artifact and contract IDs, canary scope, controls, metrics, reconciliation, approvals, and rollback receipt. Identify assumptions, owners, expirations, and evidence gaps explicitly.
Exercise disabled mode, revoked secret, egress denial, partial delivery, process restart, and rollback. Record expected and observed results, including fail-closed behavior.