Superpowered Toolkit
5 AI-powered plugins for Claude Code — one per Superpowered Professional dimension. Replace generic development workflows with methodology-grounded alternatives.
/brainstorm → /plan → /work → /review → /compound
Perception Intelligence Creation Intelligence Knowledge
Install
# Add the marketplace (once)
/plugin marketplace add aibilitycz/superpowered-toolkit
# Install individual plugins
/plugin install super-perception@superpowered-toolkit
/plugin install super-intelligence@superpowered-toolkit
/plugin install super-knowledge@superpowered-toolkit
/plugin install super-creation@superpowered-toolkit
/plugin install super-integration@superpowered-toolkit
Install what you need. Each plugin works independently — super-integration ties them together but isn't required.
Plugins
Skill: /brainstorm — Reframe problems, explore approaches, surface past work before building. Uses Socratic verification (CoVe) to challenge assumptions before accepting problem framing.
Agent: researcher — Scans codebase, docs, past brainstorms, and solutions to gather context.
Knowledge: discovery patterns
Skills:
/plan — Auto-calibrated planning (concise/standard/detailed based on scope). Surfaces past plans and adapts autonomy level to context./review — Focused code, document, or architecture review with evidence-grounded challenge (CoVe verification on each finding)/think — Expert panel simulation, devil's advocate, adversarial verification with provisional confidence/investigate — Detective-style root cause analysis (Sherlock/Poirot/Columbo lenses). Consults past solutions before investigating.
Agents: strategic-reviewer, investigator, python-reviewer, typescript-reviewer, security-reviewer, infra-reviewer, performance-reviewer
Knowledge: decision frameworks, critical evaluation, strategic decomposition, socratic patterns, infrastructure ops, security review, Python/FastAPI patterns, TypeScript/Next.js patterns, observability
Skills:
/compound — Capture solutions, context docs, learnings, and principles. Intelligent dedup — checks for existing solutions before creating new ones./psychometric-advisor — Expert guidance across 9 domains (IRT, reliability, validity, item design, scoring, AI assessment, statistics, I/O psychology, ethics)/item-designer — 5-step guided workflow for assessment items. Challenges format selection with psychometric tradeoffs./scoring-reviewer — Validate scoring changes against psychometric standards. Technical checks run autonomously; fairness checks always escalate.
Agents: knowledge-architect, psychometry-expert, assessment-auditor, skill-reviewer
Knowledge: context engineering, compounding patterns, active memory integration, 68+ peer-reviewed psychometric citations
Skills:
/work — Execute a plan from start to ship. Read tasks, implement, test, commit, quality checks, push. Defaults to autonomous mode when a plan file is provided./audio — Generate audio content — text-to-speech, podcasts, voice cloning, sound effects, dubbing, and audio isolation (ElevenLabs).
Knowledge: the 90/10 craft (AI does 90%, you craft the 10% that matters), CI/CD patterns
Skill: /orchestrate — Detects where you are in the workflow, suggests the next phase, maintains flow between plugins. Memory-informed routing — surfaces existing brainstorms and plans.
Knowledge: workflow patterns (full loop, partial flows, transitions), autonomy modes (confidence-gated escalation)
Not Every Task Needs the Full Loop
| Task | Flow |
|---|
| Quick bug fix | /investigate → /work |
| New feature | /brainstorm → /plan → /work → /review |
| Hard decision | /think |
| Code review | /review |
| Document a fix | /compound |
| Assessment question | /psychometric-advisor or /item-designer |
| Generate audio | /audio |
| Not sure where to start | /orchestrate |
What Makes This Different
Skills aren't just prompts — they're structured workflows with:
