agentleFS
tuana-vntuana-vn/system-reverse-engineer/.github/skills/detailed-design-verification/SKILL.md

detailed-design-verification

Skill0 starsChanged 2 months agoMITView on GitHub

Independently verify one artifact detailed design against current source, verified TDD, artifact registry, requirement/gap traceability, provenance/exposure evidence, dependencies, compatibility constraints, and implementability. Use before marking a design artifact implementation-ready.

The file

Show its rules and text

We split this skill into the rules it gives an agent

19 rules. About 0 are common, 14 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈14

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Verify one detailed-design artifact independently.

    line 11
  2. Distinctive:

    Do not merely check internal consistency of the document.

    line 13
  3. Distinctive:

    all material exposure paths

    line 48
  4. Distinctive:

    Do not approve a design because the signatures "look plausible".

    line 52
  5. Distinctive:

    Do not trust the section title.

    line 81
  6. Distinctive:

    If a section labeled CURRENT contains a PROPOSED component or relationship as though it already exists, record:

    line 94
  7. Very specific:

    if the TDD remains valid and the defect is in DD representation/state labeling: DD_PARTIALLY_VERIFIED

    line 101
  8. Very specific:

    if the DD asserts materially false CURRENT behavior contradicted by source: DD_CONTRADICTED_BY_SOURCE

    line 103
  9. Very specific:

    if source cannot resolve the classification: DD_BLOCKED_SOURCE_GAP

    line 105
  10. Distinctive:

    Do not certify subjective labels by repetition.

    line 112
  11. Distinctive:

    If a qualitative/zero-risk claim lacks measurable definition or evidence, record:

    line 130
  12. Distinctive:

    If implementers would need to invent a material design decision:

    line 153
  13. Very specific:

    search model/version/config branches

    line 163
  14. Distinctive:

    search earlier conditional retrieval/mapping

    line 164
  15. Distinctive:

    search error/fallback paths

    line 165
  16. Distinctive:

    search tests contradicting the design

    line 166
  17. Very specific:

    Only DD_VERIFIED may advance to WBS generation.

    line 190
  18. Distinctive:

    Update the design artifact registry status accordingly.

    line 192
  19. Distinctive:

    Use the active workflow-declared DD verification output path under the current run root.

    line 196

≈ “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

Detailed Design Verification

Mission

Verify one detailed-design artifact independently.

Do not merely check internal consistency of the document.

Requirement
→ Gap
→ Verified TDD
→ Registry Artifact
→ Detailed Design
↕
Current Source / Runtime Evidence

1. Traceability Verification

Verify:

CheckExpected
Artifact IDexists in registry
Requirement IDsmatch registry/TDD
Gap IDsverified gaps only
TDD Decision IDsverified decisions
Existing anchorsexist and have claimed role
Proposed itemsexplicitly proposed

2. Source / Runtime Verification

Check:

  • caller/callee assumptions
  • ownership/lifecycle
  • configuration/binding
  • data/state source
  • result/error semantics
  • integration boundaries
  • all material exposure paths
  • field/data provenance
  • sibling operations/counterexamples

Do not approve a design because the signatures "look plausible".

3. Design Drift Verification

Detect:

  • new behavior not approved by TDD
  • missing TDD behavior
  • changed requirement scope
  • added validation/defaults
  • omitted MUST-NOT / non-goal behavior
  • invented shared abstraction
  • unnecessary restructuring
  • hidden migration/compatibility effects

3A. Architecture State Separation Gate

This gate is mandatory whenever the DD contains architecture/component/relationship diagrams or architecture prose.

Independently enumerate every architecture view and classify every material node/relationship as:

CURRENT
PROPOSED
UNKNOWN

Do not trust the section title.

Required checks:

CheckRequired
Every CURRENT node exists in current source/evidence100%
Every CURRENT relationship is source/evidence-backed100%
PROPOSED nodes shown in a CURRENT-only view0
PROPOSED relationships shown as CURRENT0
UNKNOWN items promoted to CURRENT0
Diagram title/state labels/prose agree100%

If a section labeled CURRENT contains a PROPOSED component or relationship as though it already exists, record:

CURRENT_PROPOSED_ARCHITECTURE_CONTAMINATION

Routing:

  • if the TDD remains valid and the defect is in DD representation/state labeling: DD_PARTIALLY_VERIFIED
  • if the DD asserts materially false CURRENT behavior contradicted by source: DD_CONTRADICTED_BY_SOURCE
  • if source cannot resolve the classification: DD_BLOCKED_SOURCE_GAP

DD_VERIFIED is forbidden while any such finding remains.

3B. Quantified Risk / Confidence Claim Gate

Do not certify subjective labels by repetition.

Claims such as:

Regression Risk: NONE
Risk: LOW
Confidence: HIGH

must have explicit measurable criteria and supporting closure evidence.

Regression Risk: NONE requires evidence that all materially affected compatibility/exposure paths were enumerated and no regression mechanism remains in the verified scope.

For constructor/signature/lifecycle changes, verify all material construction/wiring sites before accepting a zero-risk claim.

If a qualitative/zero-risk claim lacks measurable definition or evidence, record:

UNSUPPORTED_QUALITATIVE_RISK_CLAIM

and do not return DD_VERIFIED until the claim is removed, made evidence-backed, or expressed as an explicit unresolved risk.

4. Implementability Verification

The design must be concrete enough that implementation tasks can be derived without guessing.

Check:

  • exact existing source anchors
  • proposed new artifact names/locations where needed
  • interfaces/contracts
  • branch/flow behavior
  • data model/config changes
  • dependency ordering
  • error/failure semantics
  • verification expectations

If implementers would need to invent a material design decision:

DD_INCOMPLETE_IMPLEMENTATION_DETAIL

5. Adversarial Counterexample Search

Before approval:

  • search alternate callers
  • search alternate implementations
  • search model/version/config branches
  • search earlier conditional retrieval/mapping
  • search error/fallback paths
  • search tests contradicting the design

6. Verdict Matrix

DD Claim / DecisionSource/TDD EvidenceCounterevidenceVerdictRequired Correction

Verdicts:

  • VERIFIED
  • PARTIALLY_VERIFIED
  • CONTRADICTED
  • SOURCE_NOT_RESOLVED
  • OUT_OF_TDD_SCOPE

7. Gate

Overall:

  • DD_VERIFIED
  • DD_PARTIALLY_VERIFIED
  • DD_CONTRADICTED_BY_SOURCE
  • DD_BLOCKED_SOURCE_GAP

Only DD_VERIFIED may advance to WBS generation.

Update the design artifact registry status accordingly.

Output

Use the active workflow-declared DD verification output path under the current run root.

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.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

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.