Verify Project

Use this skill when the user wants a real project verification, not a document review.

The goal is to answer: "Does this project actually do what it says it does, and what remains to finish it?"

Workflow

1. Establish claimed scope.
   • Read README, AGENTS.md, CONTRIBUTING, DESIGN, PRDs, ADRs, roadmap docs, issue templates, package manifests, examples, and visible user-facing copy.
   • Identify the project type, intended users, promised workflows, core features, setup path, and success criteria.
   • Treat docs as claims to verify, not as proof.
2. Inspect the implementation.
   • Review the full source tree at a practical level: routes, entrypoints, domain logic, data layer, API boundaries, tests, scripts, config, deployment files, and generated assets.
   • Use fast searches for TODO/FIXME, skipped tests, placeholder copy, mock data, disabled checks, hardcoded secrets, missing error paths, and incomplete integrations.
   • Follow the main workflows end to end through the code.
3. Run objective checks.
   • Install dependencies only when appropriate for the repo and environment.
   • Run the smallest relevant test, lint, typecheck, build, and smoke-test commands.
   • If an app can run locally, start it and verify the primary user flows through a browser or HTTP client.
   • Capture failures exactly enough that the next contributor can reproduce them.
4. Compare claims to reality.
   • Mark each major claimed capability as Complete, Partial, Missing, or Blocked.
   • Base status on code and runtime evidence, not intent.
   • Separate product gaps from code quality gaps and operational gaps.
5. Produce the report.
   • Lead with an overall completion assessment.
   • Include evidence for every major conclusion.
   • Prioritize next steps by what blocks the project from honestly shipping or being useful.

Report Format

## Verification Report

Overall status: Complete | Mostly complete | Partial | Not yet functional | Blocked

Confidence: High | Medium | Low

### What the project claims

- Claim 1
- Claim 2

### What was verified

- Check or workflow: result
- Check or workflow: result

### Capability status

| Capability | Status | Evidence | Gap |
| --- | --- | --- | --- |
| Capability name | Complete/Partial/Missing/Blocked | File, command, or observed behavior | Remaining work |

### Findings

1. Finding title
   - Evidence:
   - Impact:
   - Recommended fix:

### Next steps

1. Highest priority step
2. Next priority step
3. Next priority step

### Verification commands

```bash
commands that were run
```

### Limits of this verification

- Anything that could not be run, accessed, or proven.

Guardrails

• Do not stop at ADRs, PRDs, or Markdown. Use them only to define claims.
• Do not mark a capability complete unless code or runtime behavior supports it.
• Do not hide skipped checks, missing dependencies, auth blockers, or environment gaps.
• Do not create issues or PRs unless the user explicitly asks.
• If the repo is too large to inspect exhaustively, state the sampling strategy and residual risk.
• Prefer concrete file paths, command outputs, HTTP responses, screenshots, or browser observations over general impressions.