playbooks

home / skills / shotaiuchi / dotclaude / wf6-verify

wf6-verify skill

/dotclaude/skills/wf6-verify

This skill verifies implementation quality by running config-driven checks, evaluating prerequisites, tests, builds, lint, and success criteria.

npx playbooks add skill shotaiuchi/dotclaude --skill wf6-verify

Review the files below or copy the command above to add this skill to your agents.

Files (1)
SKILL.md
1.8 KB
---
name: wf6-verify
description: 実装を検証
argument-hint: ""
---

**Always respond in Japanese.**

# /wf6-verify

Verify implementation quality.

## Usage

```
/wf6-verify
```

## Processing

### 1. Check Prerequisites

Get active work from state.json. Verify all plan steps are completed (`plan.current_step >= plan.total_steps`). Warn if incomplete.

### 2. Run Verification

Execute each check, using commands from `.wf/config.json` (`verify.test`, `verify.build`, `verify.lint`) with auto-detection fallback based on project files:

| Check | Config Key | Fallback Detection |
|-------|-----------|-------------------|
| Tests | `verify.test` | package.json→npm test, pytest.ini/pyproject.toml→pytest, go.mod→go test |
| Build | `verify.build` | package.json→npm run build, go.mod→go build, Cargo.toml→cargo build |
| Lint | `verify.lint` | .eslintrc→npm run lint, .prettierrc→format:check, pyproject.toml→black --check, .golangci.yml→golangci-lint |

### 3. Check Success Criteria

Compare against Success Criteria from `01_KICKOFF.md`. Mark each as OK or incomplete.

### 4. Verification Summary

Display: implementation progress, test results, build status, lint status, success criteria completion, overall PASS/FAIL.

### 5. Update state.json

- On PASS: Set `current: "wf6-verify"`, `next: "wf7-pr"`
- On FAIL: Set `current: "wf6-verify"`, `next: "wf6-verify"` (re-run after fixes)

### 6. Completion Message

- PASS: Show results, inform that `/wf7-pr` is next step
- FAIL: List failed items, suggest fixes, instruct to re-run `/wf6-verify`

## Handling Failure

If verification fails: list failed items, suggest fixes, instruct to re-run `/wf6-verify`.

## Notes

- Warning for incomplete Success Criteria
- Verification can be re-run multiple times
- PR creation is handled by `/wf7-pr` after verification passes

Overview

This skill verifies implementation quality by running automated checks and comparing results to project success criteria. It inspects tests, builds, and linting using configured commands or sensible fallbacks based on project files. The skill summarizes results and updates the workflow state to guide the next step.

How this skill works

It reads active work from state.json and confirms plan completion before starting. It runs verification commands from .wf/config.json or auto-detects appropriate commands (tests, build, lint) from common project files. Results are evaluated against Success Criteria from 01_KICKOFF.md, a summary is produced, and state.json is updated with the next workflow step.

When to use it

  • Before creating a PR to ensure code quality gates pass
  • After implementing features or fixes to validate tests, build, and lint
  • When a CI run fails locally and you need a reproducible verification step
  • When you want an authoritative check against project Success Criteria

Best practices

  • Ensure state.json reflects the correct active work and plan step before running verification
  • Define verify.test, verify.build, and verify.lint in .wf/config.json for deterministic commands
  • Keep Success Criteria in 01_KICKOFF.md up to date and specific to measurable outcomes
  • Fix listed failures and re-run the skill until all checks pass
  • Run verification locally before invoking /wf7-pr to avoid back-and-forth in the workflow

Example use cases

  • Run verification after finishing implementation to confirm all plan steps are complete and tests pass
  • Use the auto-detection fallback when working with projects lacking .wf/config.json (npm, pytest, go, cargo supported)
  • Produce a clear pass/fail summary and update workflow state to enable the PR creation step
  • Debug lint and formatting issues reported by the skill, then re-run until success

FAQ

What happens if the plan is incomplete in state.json?

The skill warns about incomplete plan steps and will flag verification accordingly; complete plan steps before expecting a PASS.

How are commands selected if .wf/config.json is missing?

The skill auto-detects commands from common files (package.json, pyproject.toml, go.mod, Cargo.toml, .eslintrc, etc.) and runs appropriate test/build/lint commands.

What does the skill do after a PASS?

It sets state.json current to wf6-verify and next to wf7-pr, and informs you that /wf7-pr is the next step.

What should I do after a FAIL?

Review the listed failed items, apply suggested fixes (tests, build errors, lint issues, or unmet success criteria), then re-run /wf6-verify.