swingerman/engineer/consistency-check
Claude Code and agent skill by swingerman/engineer
- community
Checking that the SKILL.md file is still in the repository.
Use to validate DAE artifacts for schema correctness and cross-artifact consistency. Triggers — "/engineer.consistency-check", "check consistency", "validate the artifacts", "are the specs and ACs in sync", "audit this feature".
Install
npx vskill@latest install swingerman/engineer/consistency-checkSKILL.md
consistency-check
Cross-artifact validation for DAE — the methodology's take on Speckit's /analyze. Read-only: reports inconsistencies and suggests fixes; never mutates. Any agent may run it (mechanical validation, not judgment-heavy verification). checkpoint: null.
When to use
Before major pipeline transitions and as a CI gate.
- Feature scope —
/engineer.consistency-check <slug>— one feature's artifacts - Project scope —
/engineer.consistency-check --project— project-wide invariants
Not for: fixing inconsistencies (run the suggested fix skill); within-one-artifact ambiguity (clarify); code risk (crap-analyzer).
Workflow
- Resolve + load — resolve the methodology root + manifest via
${CLAUDE_PLUGIN_ROOT}/scripts/dae_resolve.py(seereferences/resolving.md). Feature scope: load every artifact infeatures/NNN-<slug>/+CHARTER.md. Project scope: loadCHARTER.md+ everyfeature.mdfrontmatter. For identifier lookups (does symbol X actually exist in the code?), prefer LSP find-definitions / workspace-symbols when an LSP MCP capability is available; fall back to grep + Read otherwise. See${CLAUDE_PLUGIN_ROOT}/references/code-lookup.md.
1.5. Run the mechanical checks first — ${CLAUDE_PLUGIN_ROOT}/scripts/dae_ontology.py <feature-dir> (or --project). It executes every constraint that is a set operation over structured data — enumerations, functional properties, parent/child inverses, cycle detection, AC↔scenario coverage closure, Principle 7 disjointness — deterministically. Take its findings as given; do not re-derive them by reading the artifacts. They are tagged in the tables below as (ontology). See ${CLAUDE_PLUGIN_ROOT}/references/ontology.md.
- Run the judgment checks (below) — the rows the script cannot decide.
- Report — errors first, then warnings; each with location + a suggested fix (which skill to run). Do not apply fixes.
- Handoff — emit a summary.
Feature-scope checks
| Check | Severity | |
|---|---|---|
feature.md slug matches folder name (ontology) | error | |
feature.md mandatory frontmatter present; status: ready ⇒ autonomy_level set (ontology) | error | |
acs.md AC IDs unique + sequential; ac_count matches actual (ontology) | error | |
Every AC has a covering @AC-N spec scenario; no scenario tags an AC that doesn't exist (ontology) | error | |
acs.md ACs in domain language (implementation-leakage heuristic) | warning | |
acs.md ACs cover the feature.md outcome | warning | |
relevant_adrs reference ADRs that exist | error | |
spec.md and .build/spec.json agree | error | |
plan.md Charter Check: every ⚠️ deviation has a matching amendment | error | |
Validation method honored? — if feature.md declares a non-default validation_method, plan.md's Test strategy section must reference it (canary phase, dashboard names, rollback trigger, etc. as relevant) | warning | |
Verification handoffs: agent_id ≠ implementer's (Principle 7) (ontology) | error | |
tracker_ref resolves on the configured tracker | warning | |
Handoff completeness: no checkpoint marked done in progress.md lacks a complete handoff | error | |
parent_feature set ⇒ parent resolves and lists this in child_features (ontology) | error / warning |
The handoff-completeness check is the after-the-fact sweep for handoff-as-gate (Foundation Design Section 5): run ${CLAUDE_PLUGIN_ROOT}/scripts/dae_handoff.py <feature-dir> — it flags any checkpoint marked done with a missing, interrupted, or unmet-exit_criteria handoff.
Project-scope checks
| Check | Severity | |
|---|---|---|
CHARTER.md section 5 roles == manifest.team.default_roles | error | |
manifest.yml schema valid; enum values legal | error | |
| Feature folder numbering monotonic; no reuse (ontology) | warning | |
| One feature branch per feature; one feature per roadmap item (ontology) | warning | |
parent_feature chains do not cycle (ontology) | error | |
| ADR not referenced by any feature in N months (N from manifest, default 6) | warning | |
Every features/*/ has a non-empty feature.md | warning | |
team.default_roles includes verifier when independence enforced | error |
Handoff
Emit per ${CLAUDE_PLUGIN_ROOT}/references/handoff-summary.md. checkpoint: null; artifacts: []. Feature scope → handoff in that feature's handoffs/; --project scope → handoff in .engineer/handoffs/. Any errors → human_action_needed: yes. recommended_next: resolve via clarify / feature-edit / plan, then re-run.
References
${CLAUDE_PLUGIN_ROOT}/references/ontology.md— the (ontology) rows: whatdae_ontology.pydecides and why the rest stays here- Foundation Design — every schema checked here
- Tracker Integration —
tracker_refresolution
Security
- Security scanPassed, 100 of 100Pattern checks on the skill's files.
- Model reviewNot runA model review is scheduled for the next rescan.
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingengineer/skills/consistency-check/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/swingerman/engineer/consistency-check)