TaylorHuston

validate-spec

23
2
# Install this skill:
npx skills add TaylorHuston/local-life-manager --skill "validate-spec"

Install specific skill from multi-skill repository

# Description

Validate spec completeness and implementation compliance. Use before approval (--pre) or before completion (--post).

# SKILL.md


name: validate-spec
description: "Validate spec completeness and implementation compliance. Use before approval (--pre) or before completion (--post)."
model: claude-opus-4-5-20251101
allowed-tools: Read, Glob, Grep


/validate-spec

Validate spec quality (pre-implementation) or implementation compliance (post-implementation).

Usage

/validate-spec SPEC-001              # Auto-detect mode based on status
/validate-spec SPEC-001 --pre        # Pre-implementation validation
/validate-spec SPEC-001 --post       # Post-implementation compliance

Mode Detection

Spec Status Default Mode Purpose
draft, review --pre Validate before approval
approved --pre Final check before implementation
in_progress, complete --post Verify implementation matches

Pre-Implementation Validation (--pre)

1. Structure Completeness

Required sections:
βœ“ Summary (present)
βœ“ Problem Statement (present)
βœ“ Acceptance Scenarios (3 found)
βœ“ Agent Context (present)
βœ“ Out of Scope (present)
βœ— Dependencies (MISSING)

2. Acceptance Scenario Quality

Scenario 1: "User signs up"
βœ“ Has Given/When/Then clauses
βœ“ Specific, not vague

Scenario 2: "Invalid credentials"
⚠ Then clause is vague: "shows error"
  Suggest: "shows 'Invalid email or password' message"

Scenarios must be:
- Specific: Not "shows error" but "shows error X"
- Testable: Can become automated test
- Complete: Given/When/Then present

3. Agent Context Validation

Files referenced:
βœ“ src/lib/auth.ts exists
βœ— src/middleware/session.ts NOT FOUND

4. Success Metrics Measurability

Metric 1: "Page load < 2s"
βœ“ Measurable, has threshold

Metric 2: "Good user experience"
βœ— NOT MEASURABLE
  Suggest: "User satisfaction > 4/5"

5. Conciseness Check

Word count: 847 words
Guideline: < 500 words
⚠ Consider splitting into multiple specs

6. Open Questions Check

2 open questions found:
1. "Which OAuth providers?"
2. "Session timeout?"

⚠ Resolve before 'approved' status

Pre-Validation Summary

Ready for approval: NO

Fix:
- Add missing Dependencies section
- Refine vague scenario
- Make metrics measurable
- Resolve open questions

Post-Implementation Validation (--post)

1. Acceptance Scenario Coverage

Scenario: "User signs up"
βœ“ Found: auth/signup.test.ts:15

Scenario: "Session expires"
βœ— NO TEST FOUND

Coverage: 2/3 scenarios (67%)

2. Success Metrics Verification

Metric: "Page load < 2s"
  Result: 1.4s average
  βœ“ PASS

Metric: "95% coverage"
  Result: 97.2%
  βœ“ PASS

3. Out-of-Scope Violations

Out of scope: "Social login", "2FA"

Scanning...
βœ“ No oauth/social code found
⚠ Found "password strength" - potential violation

4. Agent Constraints Compliance

Constraint: "No new dependencies"
βœ“ package.json unchanged

Constraint: "Files outside scope"
⚠ src/utils/helpers.ts modified

5. Definition of Done

- [x] Acceptance scenarios tested
- [ ] No console errors ← incomplete

Post-Validation Summary

Implementation compliance: PARTIAL

Issues to resolve:
1. Add test for "Session expires"
2. Review password complexity code
3. Fix console warnings

Interactive Fixes

When issues found:

AI: Let's refine Scenario 2...
    What should the error message say?

User: "Invalid email or password"

AI: Updated. Now the metrics...
    Remove "Good UX" or replace with measurable?

User: Remove it

AI: Done. Re-validating...

    ## Validation: PASS
    Ready to move to 'review' status?

Integration

/spec β†’ /validate-spec --pre β†’ [approval] β†’ /plan β†’ /implement β†’ /validate-spec --post β†’ /complete

# Supported AI Coding Agents

This skill is compatible with the SKILL.md standard and works with all major AI coding agents:

Learn more about the SKILL.md standard and how to use these skills with your preferred AI coding agent.