skills/superpowers-integrator/SKILL.md
Use this skill whenever you need to audit or upgrade a custom skill's integration with the superpowers plugin workflow. ALWAYS trigger on: "integrate with superpowers", "upgrade skill integration", "audit my skill", "add superpowers workflow to X", "make this skill superpowers-aware", "check skill integration", "re-sync superpowers workflow", "refresh superpowers integration", "superpowers compliance check". Implicit triggers: user just created a new custom skill and wants it aligned with the superpowers development workflow; user wants to know whether an existing skill correctly routes through brainstorming/plans/TDD/verification/review; superpowers ships a new skill and existing custom skills need re-audit; user is maintaining a skills pack and wants a single source of truth for integration patterns. This is a META-skill: it reads other skills' SKILL.md files and produces either a findings report or an applied upgrade. It does NOT generate domain code. It is IDEMPOTENT — safe to re-run against the same skill at any time. The taxonomy (`references/skill-class-taxonomy.md`) and block templates (`templates/blocks.md`) are the single source of truth for "how custom skills integrate with superpowers". When superpowers adds a new skill, the user's learned a better pattern, or the skill has drifted, update the source-of-truth files and re-run this integrator against every custom skill — one command, full pack re-synced.
npx skillsauth add OmexIT/claude-skills-pack superpowers-integratorInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Audits and upgrades custom skills to integrate with the superpowers plugin workflow. Re-runnable. Single source of truth lives in references/skill-class-taxonomy.md + templates/blocks.md — update those to change the pattern for every custom skill in one shot.
Parse $ARGUMENTS:
fintech-ledger): look up under skills/<name>/SKILL.mdskills/temporal-workflow/SKILL.mdskills/*/ → run phases 1-5 for each target--audit-only flag: run phases 1-3 only, skip applicationFor each target, run phases 1-5 in order.
Report per-target outcome:
──────────────────────────────────────────────
TARGET: <skill-name>
Classification: <class>
Audit score: <current>/10 → <proposed>/10
Status: <no-change | upgrade-proposed | upgrade-applied | error>
──────────────────────────────────────────────
Read the target SKILL.md. Apply heuristics from references/skill-class-taxonomy.md in priority order (specific first):
ledger|payment|wallet|settlement|saga|escrow|balance OR description mentions BigDecimal/double-entry/reconciliation/pgledger/blnk → class = code-generator-money-criticalmigration|schema|changelog|liquibase|flyway|ddl OR output contract produces .sql files → class = code-generator-sqlreviewerdebuggerrefactorplannercode-generatorIf a skill fits multiple classes, always pick the most specific (money-critical > sql > code-generator).
If a skill genuinely spans classes, report AMBIGUOUS and ask the user to confirm.
Report classification:
🔍 CLASSIFIED — <skill-name>
Class: <class>
Reason: <which heuristic matched>
Evidence:
- name contains: <match>
- description mentions: <match>
- output contract: <match>
Load references/integration-checklist.md. Apply the universal checks (U1-U5) plus the class-specific checks to the target SKILL.md.
For each check:
Compute score: (passed_items / total_items) * 10, rounded.
Rule: any skill scoring < 9 needs upgrade. Skills scoring 9-10 are compliant; report and move on without changes.
Report:
📋 AUDIT — <skill-name> [<class>]
Universal checks:
U1 ✅ Has "Before You Start" H2 section
U2 ❌ Section is positioned wrong (should be after title, before section 0)
U3 ⚠️ References wrong superpowers skills for this class
U4 ❌ handoff field doesn't mention verification-before-completion
U5 ✅ Anti-patterns section present
Class-specific checks (<class>):
<CG1, CG2, ...>
Score: <n>/10 (needs upgrade)
If score ≥ 9, skip to phase 5 (report no-change).
Otherwise, load templates/blocks.md and extract the block matching the classified class. Substitute these placeholders:
| Placeholder | Source |
|---|---|
| {SKILL_NAME} | Target skill's name: frontmatter field |
| {SKILL_PURPOSE} | One-line purpose extracted from the skill's first paragraph or description sentence |
| {DOMAIN_RIGOR_NOTE} | Optional additional rigor note — leave empty unless the target skill has domain-specific hard rules worth preserving |
Then compute the diff:
Insert (always):
# Title heading and before section ## 0. (or the first ## section)Modify (if any of these fail):
handoff: field — append reference to superpowers:verification-before-completion and superpowers:requesting-code-review as appropriate for the classsuperpowers:systematic-debugging if missingShow the proposed diff to the user:
📝 PROPOSED UPGRADE — <skill-name>
INSERT after line <n> (after title, before section 0):
──────────────────────────────────────────
## Before You Start — Superpowers Workflow
<rendered block>
──────────────────────────────────────────
MODIFY line <n> (handoff):
- handoff: "<current>"
+ handoff: "<current + superpowers chain>"
Apply? [yes/no/audit-only]
On user confirmation (yes):
handoff: line if neededOn audit-only or no: report findings without applying and move on.
Report:
✅ UPGRADED — <skill-name>
Class: <class>
Score: <old> → <new>
Files: skills/<name>/SKILL.md
Changes:
- Inserted "Before You Start" block (<n> lines)
- Modified handoff field
This skill is idempotent. Safe to re-run at any time. Use cases:
| Trigger | Action |
|---|---|
| New custom skill added | /superpowers-integrator <new-skill> |
| superpowers ships a new skill | Update references/skill-class-taxonomy.md and templates/blocks.md, then /superpowers-integrator skills/*/ |
| User learns a better pattern | Update templates/blocks.md, then re-run against affected classes |
| Monthly maintenance check | /superpowers-integrator skills/*/ --audit-only to identify drift without applying |
| Single skill drifted | /superpowers-integrator <skill> to re-sync |
The intended workflow for keeping custom skills in sync with superpowers:
┌────────────────────────────────────────────────────┐
│ 1. Notice: a skill feels stale or superpowers new │
└──────────────────┬─────────────────────────────────┘
▼
┌────────────────────────────────────────────────────┐
│ 2. Edit the source of truth: │
│ - references/skill-class-taxonomy.md │
│ - templates/blocks.md │
│ - references/integration-checklist.md │
└──────────────────┬─────────────────────────────────┘
▼
┌────────────────────────────────────────────────────┐
│ 3. Re-audit the pack: │
│ /superpowers-integrator skills/*/ --audit-only │
└──────────────────┬─────────────────────────────────┘
▼
┌────────────────────────────────────────────────────┐
│ 4. Review the findings, then apply: │
│ /superpowers-integrator skills/*/ │
└──────────────────┬─────────────────────────────────┘
▼
┌────────────────────────────────────────────────────┐
│ 5. Commit the pack + re-install to ~/.claude/ │
└────────────────────────────────────────────────────┘
One source of truth. One command to re-sync. No drift.
produces:
- type: "audit-report"
format: "markdown"
path: "claudedocs/superpowers-audit-<timestamp>.md"
sections: [per_skill_audit, pack_summary, proposed_upgrades]
- type: "skill-upgrade"
format: "markdown-edit"
paths: ["skills/<name>/SKILL.md"]
handoff: "Write claudedocs/handoff-superpowers-integrator-<timestamp>.yaml — suggest: verify (run a sample skill invocation to confirm upgraded workflow), superpowers:requesting-code-review (for the pack if committed to git)"
| File | When to read |
|---|---|
| references/skill-class-taxonomy.md | Classifying a skill, adding a new class, updating class workflows |
| references/integration-checklist.md | Running the audit phase |
| templates/blocks.md | Rendering the "Before You Start" block for a classified skill |
--apply-all is explicitly passed).tools
Use this skill to verify a completed implementation through live testing — API calls, database state checks, and UI automation with Playwright. Triggers include: "test the implementation", "verify this works", "run API tests", "check the database", "test the UI", "end-to-end verify", "smoke test", "sanity check the implementation", "manually test", or any time an implementation needs post-build validation beyond unit tests. Also triggered automatically by spec-to-impl during the integration review phase. Use this when you want real evidence the system works — not just that tests compile. Can consume a pre-generated e2e/test-plan.yaml from spec-to-impl for fully automated test execution.
development
--- name: ux-review description: Evaluate a UI/UX design or implementation using heuristic analysis, accessibility audit, and cognitive walkthrough. Triggers: "UX review", "usability review", "heuristic evaluation", "accessibility audit", "is this usable". argument-hint: "[feature / screen / URL / mockup]" effort: high --- # UX review ## What I'll do Evaluate a design or implementation for usability, accessibility, and user experience quality using established heuristic frameworks. ## Inputs
development
--- name: user-flow description: Map user journeys through a feature or product, identifying key paths, decision points, friction, error states, and edge cases. Triggers: "user flow", "user journey", "flow diagram", "happy path", "user path". argument-hint: "[feature / user goal]" effort: medium --- # User flow ## What I'll do Map the complete user journey for a feature — from entry point through completion — including happy paths, error states, edge cases, and decision points. > **user-flow
development
Use this skill to produce complete UI/UX design artifacts from a specification document or panel analysis. Triggers include: "design the UI for this spec", "create wireframes", "design this panel", "UX design from spec", "generate component specs", "design tokens", "create the UI design for", "design system for", "wireframe this feature", "design a UI", "create a design system", "design this component", "design the layout", "create a style guide", "design a screen", "UI/UX review", "typography system", "color system", "spacing system", "design this feature", "design the dashboard", "design the onboarding", "create a component library", "design review", "audit the design", "improve the UI", "redesign this", "design system documentation", "create design guidelines", "responsive design", "mobile design", "dark mode design", "design the brand", or any time a spec/panel analysis document needs to be transformed into actionable UI/UX deliverables before implementation. Also triggers for standalone design system creation, component design, design reviews, dark mode/responsive variants, and developer handoff — even before code is involved. Orchestrates a multi-agent design team (UX Lead, UI Designer, Component Architect, Accessibility Reviewer, Design System Engineer, Design Reviewer) in parallel waves. Outputs feed directly into spec-to-impl's FE agent and figma-to-code.