brainstorming
Structured design dialogue that validates ideas before implementation begins.
criteria-audit
Validate EXECUTION_PLAN.md for verification metadata, manual reasons, and testability. Use when preparing Phase 1 or after editing EXECUTION_PLAN.md.
Full skill instructions
Audit EXECUTION_PLAN.md to ensure acceptance criteria are automation-ready and use the verification metadata format.
$1 (optional) = directory containing EXECUTION_PLAN.md
$1/EXECUTION_PLAN.mdEXECUTION_PLAN.md from the current working directoryCopy this checklist and track progress:
Criteria Audit Progress:
- [ ] Step 1: Read EXECUTION_PLAN.md
- [ ] Step 2: Parse phases, tasks, and acceptance criteria
- [ ] Step 3: Validate verification metadata
- [ ] Step 4: Report issues and summarize
Resolve the plan path: if $1 is provided, use $1/EXECUTION_PLAN.md; otherwise use EXECUTION_PLAN.md in the current working directory.
For each task, collect:
(TEST))Verify: lineReason: lineAlso collect Pre-Phase Setup items and their Verify: lines.
(TEST), (CODE), (LINT),
(TYPE), (BUILD), (SECURITY), (BROWSER:DOM), (MANUAL), (MANUAL:DEFER) etc.Verify: line unless it is MANUAL or MANUAL:DEFER.MANUAL and MANUAL:DEFER criteria must include a Reason: line.Verify: command.Read ~/.claude/skills/auto-verify/PATTERNS.md for the full pattern matching table
and MANUAL decision tree. If PATTERNS.md is not found, skip the false-MANUAL check and note the limitation in the report output (e.g., "PATTERNS.md not found — false-MANUAL detection skipped").
For each criterion tagged (MANUAL), check if it contains keywords from the
Pattern Matching Table that indicate it CAN be automated (priorities 1-10).
If it matches any automatable pattern, it is a false MANUAL tag.
Only criteria matching the "Truly Manual Patterns" section (subjective UX/brand/tone judgment) should remain as MANUAL.
For each (MANUAL) criterion (not MANUAL:DEFER):
(MANUAL:DEFER)
→ Reason: "No downstream dependency detected"For each (MANUAL:DEFER) criterion:
MANUAL (blocking) — downstream dependency exists"Provide a structured report:
CRITERIA AUDIT
==============
Tasks Checked: {N}
Criteria Checked: {N}
Issues Found: {N}
Missing Type Tags:
- Task 1.2.A: "{criterion}"
Missing Verify Lines:
- Task 1.3.B: "{criterion}"
Manual Missing Reason:
- Task 2.1.A: "{criterion}"
Pre-Phase Setup Missing Verify:
- Phase 1: "{setup item}"
False MANUAL Tags (should be automated):
- Task 1.2.A: (MANUAL) "{criterion}"
→ Suggest: (CODE) — Verify: `curl -sf {url} -o /dev/null`
→ Reason: Contains "endpoint"/"returns" — automatable via curl
- Task 2.1.B: (MANUAL) "{criterion}"
→ Suggest: (BROWSER:DOM) — Verify: route=`/page`, selector=`.class`
→ Reason: Contains "visible"/"displays" — automatable via browser
MANUAL Summary:
Total MANUAL criteria: {N}
Likely false tags: {N} (should be retagged to automated)
Truly manual: {N} (subjective judgment)
DEFER Classification:
Total MANUAL: {N} blocking, {M} deferrable
Suggested retags: {list of MANUAL → MANUAL:DEFER or vice versa}
Status: PASS | WARN | FAIL
FAIL if any false MANUAL tags are found. WARN if MANUAL criteria exceed 10% of total criteria. PASS otherwise.
| Situation | Action |
|---|---|
| EXECUTION_PLAN.md not found at the resolved path | Report "EXECUTION_PLAN.md not found" with the path checked and stop |
PATTERNS.md not found (~/.claude/skills/auto-verify/PATTERNS.md) | Skip false-MANUAL detection, note limitation in report output |
| Plan file is empty or has no parseable phases/tasks | Report "No phases or tasks found" and mark audit as NOT APPLICABLE |
| A criterion has multiple conflicting type tags | Flag as ambiguous, list all tags found, and recommend the user pick one |
| Plan file exceeds 2000 lines | Process in chunks, warn user that partial analysis may miss cross-phase dependencies |
Verify: line.Structured design dialogue that validates ideas before implementation begins.
Comprehensive design intelligence for web and mobile UI/UX across 10 technology stacks.
Comprehensive implementation plans for multi-step tasks, breaking down specs into bite-sized, testable steps.
Introduction to the obra skills system with mandatory skill invocation rules and best practices.
Execute a written implementation plan with critical review and task checkpoints.
Delegate independent tasks to specialized agents working concurrently with isolated context.
Isolated git worktrees with smart directory selection and safety verification.
Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.
Plan searchable and shareable content that drives traffic, builds authority, and generates leads.
README-first repository scanner that extracts commands and classifies reproduction candidates without executing them.
Brainstorm and prioritize marketing strategies tailored to your SaaS stage, budget, and goals.
Plan and optimize your website's page hierarchy, navigation, URL structure, and internal linking.
The world's fastest calendar for remote work
Capture, organize, and utilize your knowledge effortlessly.
Instant Summaries of Audio & Video Interviews with AnySummary
Transform PDFs into engaging mind maps.
Chat with any PDF instantly
Create an optimal daily plan using your voice
A productivity platform to centralize organizational knowledge and workflows with contextual AI assistance.
AskYourPDF Pricing Plans: Tailored to Your Needs