agentleFS
ArvorCoArvorCo/Relentless/.claude/skills/checklist/SKILL.md

checklist

Skill98 starsChanged 9 months agoMITView on GitHub

Generate quality validation checklist from spec, plan, and tasks. Use after creating tasks. Triggers on: create checklist, quality checklist, validation checklist.

We split this skill into the rules it gives an agent

9 rules. 0 are common, 9 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 8 edited versions we read keep these.

  • 9

    Distinctive

    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:

    Every generated checklist MUST include these testing items:

    line 29
  2. Distinctive:

    Every story in the checklist MUST include:

    line 45
  3. Distinctive:

    Every checklist MUST include:

    line 60
  4. Distinctive:

    Read spec.md, plan.md, and tasks.md

    line 73
  5. Distinctive:

    Verify these files exist:

    line 87
  6. Distinctive:

    Create 7-10 categories based on feature domain:

    line 105
  7. Distinctive:

    If you have any doubts or suggestions about the checklist, please, interview the user.

    line 121
  8. Distinctive:

    Ensure checklist includes items for constitution MUST rules:

    line 146
  9. Distinctive:

    If validation fails, fix the errors and re-run

    line 162

“Common” is measured: kept by at least 6 in 10 of the 8 edited versions of this file we could read. “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

Quality Validation Checklist Generator

Generate domain-specific quality checklists to validate implementation completeness.


SpecKit Workflow

This skill runs after tasks, before analyze:

specify → plan → tasks → checklist → analyze → implement

Purpose:

  • Generate validation checklist from spec/plan/tasks
  • Ensure quality gates are defined
  • Ensure TDD requirements are included
  • Ensure routing compliance is checked
  • Provide implement phase with verification criteria

TDD is MANDATORY

Every generated checklist MUST include these testing items:

Testing & TDD (Required for ALL stories)

  • Unit tests written BEFORE implementation
  • Unit tests verified to FAIL before coding
  • Integration tests for API/data flows
  • E2E tests for user-facing flows
  • Test coverage meets 80% target
  • All tests pass

These items are non-negotiable and must appear in every checklist.


Mandatory Quality Gates

Every story in the checklist MUST include:

Quality Gates (Non-Negotiable)

  • bun run typecheck passes with 0 errors
  • bun run lint passes with 0 errors AND 0 warnings
  • bun test passes
  • No debug code (console.log, debugger)
  • No unused imports or variables

All three commands must pass - no exceptions.


Routing Validation Items

Every checklist MUST include:

Routing Compliance

  • Routing preference documented in spec.md
  • prd.json has routing metadata for all stories
  • Complexity classifications match story scope
  • Estimated costs are reasonable
  • Mode selection (free/cheap/good/genius) is appropriate

The Job

  1. Read spec.md, plan.md, and tasks.md
  2. Extract quality requirements from constitution
  3. Include mandatory TDD items
  4. Include mandatory quality gate items
  5. Include routing compliance items
  6. Generate comprehensive checklist
  7. Save to checklist.md

Important: Checklist items get merged into prd.json acceptance criteria!


Step 1: Locate Feature Files

Verify these files exist:

  • relentless/features/NNN-feature/spec.md
  • relentless/features/NNN-feature/plan.md
  • relentless/features/NNN-feature/tasks.md

Step 2: Load Context

Read:

  1. relentless/constitution.md - Quality standards and requirements
  2. Feature files (spec, plan, tasks)
  3. Identify the feature domain (auth, payments, dashboard, etc.)

Step 3: Generate Checklist Categories

Create 7-10 categories based on feature domain:

Mandatory Categories (always include):

  • 0. Quality Gates - typecheck, lint, test
  • 1. TDD Compliance - tests before code
  • 2. Routing Compliance - metadata validation

Common Categories (include as needed):

  • Schema & Database
  • Backend Logic
  • Frontend Components (if applicable)
  • API Integration
  • Security & Permissions
  • Performance & UX
  • Documentation

If you have any doubts or suggestions about the checklist, please, interview the user.

Step 4: Generate Checklist Items

For each category, create 5-10 specific validation items:

Format:

- [ ] CHK-001 [US-001] Specific, testable requirement
- [ ] CHK-002 [Gap] Missing specification identified
- [ ] CHK-003 [Ambiguity] Unclear requirement needs clarification
- [ ] CHK-004 [Edge Case] Potential edge case to handle

Guidelines:

  • 80% of items should reference specific user stories [US-XXX]
  • 20% should identify gaps, ambiguities, or edge cases
  • Each item must be specific and testable
  • Total: 30-50 items across all categories

Step 5: Reference Constitution

Ensure checklist includes items for constitution MUST rules:

  • Type safety requirements (no any types)
  • Testing coverage (80% minimum)
  • Security standards
  • Performance expectations
  • Code quality standards (0 lint warnings)

Step 6: Save & Validate

  1. Save to relentless/features/NNN-feature/checklist.md
  2. Run the validator to ensure checklist.md is correctly formatted:
    .claude/skills/validators/scripts/validate-checklist.sh "relentless/features/NNN-feature/checklist.md"
    • If validation fails, fix the errors and re-run
    • Warnings are acceptable but should be reviewed
  3. Report:
    • Total checklist items: N
    • Items per category
    • Mandatory items included: TDD ✓, Quality Gates ✓, Routing ✓
    • Constitution compliance items: N
    • Gaps identified: N
    • Validation: PASS/FAIL
    • Next step: /relentless.convert (if not done) or /relentless.analyze

Example Checklist

# Quality Checklist: User Authentication

**Purpose:** Validate completeness of user authentication implementation
**Created:** 2026-01-11
**Feature:** [spec.md](./spec.md)

## 0. Quality Gates (MANDATORY - Every Story)

- [ ] CHK-001 `bun run typecheck` passes with 0 errors
- [ ] CHK-002 `bun run lint` passes with 0 errors AND 0 warnings
- [ ] CHK-003 `bun test` passes
- [ ] CHK-004 No debug code (console.log, debugger statements)
- [ ] CHK-005 No unused imports or variables

## 1. TDD Compliance (MANDATORY)

- [ ] CHK-006 Tests written BEFORE implementation code
- [ ] CHK-007 Tests verified to FAIL before writing implementation
- [ ] CHK-008 Unit test coverage ≥80%
- [ ] CHK-009 Integration tests for all API endpoints
- [ ] CHK-010 E2E tests for complete user flows

## 2. Routing Compliance (MANDATORY)

- [ ] CHK-011 Routing preference documented in spec.md
- [ ] CHK-012 prd.json has routing metadata for all stories
- [ ] CHK-013 Complexity classifications correct for story scope
- [ ] CHK-014 Estimated costs are reasonable
- [ ] CHK-015 Mode selection appropriate for complexity

## 3. Schema & Database

- [ ] CHK-016 [US-001] User table has all required fields (id, email, password_hash, confirmed, timestamps)
- [ ] CHK-017 [US-001] Email field has UNIQUE constraint
- [ ] CHK-018 [US-001] Indexes created on frequently queried fields (email)
- [ ] CHK-019 [Gap] Consider adding last_login_at timestamp for analytics
- [ ] CHK-020 [US-001] Migration script tested on clean database

## 4. Backend Logic

- [ ] CHK-021 [US-001] Password hashing uses bcrypt with appropriate cost factor
- [ ] CHK-022 [US-001] Email validation prevents common typos and invalid formats
- [ ] CHK-023 [US-002] JWT tokens include user ID and expiration
- [ ] CHK-024 [US-002] Token verification handles expired tokens gracefully
- [ ] CHK-025 [Ambiguity] Password reset flow not specified - out of scope?

## 5. API Integration

- [ ] CHK-026 [US-001] POST /api/auth/register returns 201 on success
- [ ] CHK-027 [US-001] Register endpoint returns 400 for invalid email
- [ ] CHK-028 [US-001] Register endpoint returns 409 for duplicate email
- [ ] CHK-029 [US-002] POST /api/auth/login returns 401 for wrong password
- [ ] CHK-030 [US-002] Login endpoint returns 403 for unconfirmed account
- [ ] CHK-031 [Edge Case] Rate limiting on auth endpoints to prevent brute force

## 6. Testing & Validation

- [ ] CHK-032 [US-001] Unit tests for password hashing utility
- [ ] CHK-033 [US-001] Unit tests for email validation
- [ ] CHK-034 [US-002] Integration test for full registration flow
- [ ] CHK-035 [US-002] Integration test for login flow
- [ ] CHK-036 [US-003] E2E test for email confirmation
- [ ] CHK-037 [Constitution] Test coverage meets minimum 80% requirement

## 7. Security & Permissions

- [ ] CHK-038 [US-001] Passwords never logged or exposed in errors
- [ ] CHK-039 [US-001] Password requirements enforced (min length, complexity)
- [ ] CHK-040 [US-002] JWT secret stored in environment variable, not code
- [ ] CHK-041 [US-002] Token expiration validated on every request
- [ ] CHK-042 [Gap] Consider adding account lockout after N failed attempts

## 8. Performance & UX

- [ ] CHK-043 [US-001] Registration completes within 2 seconds
- [ ] CHK-044 [US-002] Login completes within 1 second
- [ ] CHK-045 [US-003] Confirmation email sent within 30 seconds
- [ ] CHK-046 [Edge Case] Graceful handling when email service is down

## 9. Documentation

- [ ] CHK-047 API endpoints documented with examples
- [ ] CHK-048 Environment variables documented in README
- [ ] CHK-049 Database schema documented

Notes

  • Checklist items will be merged into acceptance criteria during convert
  • Reference user stories with [US-XXX] tags
  • Identify gaps and ambiguities with [Gap] and [Ambiguity] tags
  • Each item must be independently verifiable
  • Balance between comprehensiveness and practicality
  • Mandatory sections (Quality Gates, TDD, Routing) must always be included

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.