Skip to content

Implementation and Verification

Scale the workflow to risk while always preserving evidence.

Change sizing

For a small local change:

  1. Inspect the implementation, callers, tests, and configuration.
  2. Make the narrowest coherent edit.
  3. Run focused tests and the nearest static checks.

For a public API, persistence, authentication, concurrency, deployment, migration, or cross-module change:

  1. Map consumers and compatibility constraints.
  2. State failure, rollback, and migration behavior.
  3. Add contract, integration, or end-to-end coverage as appropriate.
  4. Run broader checks and a realistic smoke test when practical.

Test selection

RiskMinimum useful evidence
Pure deterministic logicUnit tests with boundaries and invalid input
Public API or libraryContract tests, types/compilation, realistic usage examples
Adapter or persistenceIntegration test against realistic semantics
Authorization or sensitive dataAllowed and denied cases, ownership checks, non-disclosure
Concurrency or retriesDuplicate, timeout, cancellation, ordering, and cleanup cases
Migration or deploymentCompatibility sequence, dry run or staging evidence, success and rollback signals

Tests should validate observable behavior. Avoid coupling them to private methods or incidental object structure.

Verification order

Run fast, focused checks first so failures are attributable. Then expand according to blast radius:

  1. Targeted tests.
  2. Type checking or compilation.
  3. Lint and formatting verification.
  4. Related package or module tests.
  5. Full build or suite when justified.
  6. Runtime or integration smoke checks.

Use project-provided commands. Check scripts before running them and avoid fixing, snapshot-updating, migration, deployment, or destructive options unless they are in scope.

Failure handling

When a check fails:

  • determine whether the change caused it;
  • fix in-scope regressions;
  • do not rewrite unrelated failing areas;
  • report pre-existing or environment-blocked failures precisely;
  • never disable a meaningful check to obtain a green result.

Completion evidence

The final report should identify executed commands, outcomes, anything not run, and why. “Looks correct” is not verification.


Canonical source: skills/nestjs-professional-software-engineering/references/implementation-verification.md. This page is generated during the documentation build.

Open-source guidance for deliberate NestJS engineering.