superpowered-toolkit

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

super-perception — Discovery

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

super-intelligence — Strategic Thinking

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

super-knowledge — Knowledge & Measurement

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

super-creation — Execution

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

super-integration — Orchestration

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: