agentleFS
nesnilnehcnesnilnehc/ai-cortex/skills/review-implementation-alignment/SKILL.md

review-implementation-alignment

Skill8 starsChanged 2 months agoMITView on GitHub

Compare an implemented change and its verification evidence with approved requirements, designs and tasks. Post-coding atomic functional-alignment review; outputs findings without repairing.

The file

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    After implementation and before declaring functional completion

    line 28
  2. Distinctive:

    After repair when acceptance or design alignment may have changed

    line 29
  3. Distinctive:

    Follow contracts and fields across layers and confirm production composition reaches delivered behavior.

    line 37
  4. Distinctive:

    Do not infer an unrecorded requirement or approve scope changes.

    line 47
  5. Distinctive:

    Do not rewrite artifacts or code.

    line 48
  6. Distinctive:

    Do not substitute green tests for acceptance traceability.

    line 49
  7. Distinctive:

    Do not fold engineering-quality findings into this category.

    line 50

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Skill: Review Implementation Alignment

Purpose

Evaluate an implemented change against implementation-alignment-quality. This is an atomic post-coding functional-alignment review. It is not a meta-skill, code-quality orchestrator or repair loop.

Core objective

Find omissions, scope drift, contract/data divergence, missing production wiring and unsupported completion claims that a file-by-file review or implementation-shaped test suite can miss.

Scope boundaries

This Skill compares approved artifacts, code and evidence. Intrinsic engineering quality belongs to orchestrate-code-review; test execution belongs to automate-tests; fixes and repetition belong to orchestrate-repair-loop.

Two neighbouring cases are easy to file here by mistake. Delivered code that nothing reaches is this Skill's: ALN-003 requires each in-scope design element to have a reachable production implementation. The implementation a change replaced and left behind is not: it belongs to ARC-010 of architecture-quality, which review-architecture runs over the same scope in the sibling engineering gate. Name it and move on rather than emitting a finding for it.

Use cases

  • After implementation and before declaring functional completion
  • After repair when acceptance or design alignment may have changed
  • For a change whose tests are green but an approved item may be absent or unwired

Behavior

  1. Load the alignment Rule and the authoritative artifact chain in order: requirement, functional design when present, technical design, tasks.
  2. Reject superseded or draft artifacts as completion authority. Resolve the exact implementation diff/scope and available test/manual evidence.
  3. Build a traceability matrix from each in-scope acceptance/design/task item to production location and verification evidence.
  4. Follow contracts and fields across layers and confirm production composition reaches delivered behavior.
  5. Apply ALN items independently and emit category cognitive-alignment, citing fully qualified Rule IDs.
  6. Return Rule coverage using the exact passed, waived, not_applicable and evidence_limited fields from the findings-list Spec. Missing authoritative artifacts are evidence limitations, not invented requirements.

Input and output

Input is the approved artifact paths, implementation scope and evidence. Output follows findings-list.

Restrictions

  • Do not infer an unrecorded requirement or approve scope changes.
  • Do not rewrite artifacts or code.
  • Do not substitute green tests for acceptance traceability.
  • Do not fold engineering-quality findings into this category.

Self-Check

  • Authoritative artifacts and statuses were resolved.
  • Every in-scope acceptance item maps to code and independent evidence.
  • Contracts, data fields, tactics, production wiring and task completion were checked where applicable.
  • Every finding uses category cognitive-alignment and cites an ALN Rule ID.
  • Engineering review and repair were not performed here.

Examples

Example 1: acceptance item omitted

The design requires timeout and caching; code implements only timeout and tests mirror the code. Emit ALN-001 and ALN-007 findings at the acceptance/test locations.

Example 2: field disappears between layers

The repository maps a new field, but the API response omits it. Emit ALN-004 with the contract and mapping locations.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.