Claude Skills Library & SKILL.md Templates

---
name: code-review-partner
description: Turn Claude into a senior reviewer that prioritizes defects, regressions, missing tests, and maintainability risks.
---

# Code Review Partner

## When To Use
Use when reviewing a pull request, diff, or changed files before merge.

## What Claude Should Optimize For
- Finds high-signal bugs first
- Separates blockers from polish
- Suggests concrete fixes

## Workflow
1. Read the diff and surrounding code.
2. Identify behavior changes and risk areas.
3. Check tests and edge cases.
4. Return findings ordered by severity.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- code review
- pull requests
- bugs
- tests

---
name: debugging-detective
description: Guide Claude through systematic bug reproduction, hypothesis testing, log analysis, and minimal fixes.
---

# Debugging Detective

## When To Use
Use when an error, failing test, flaky behavior, or production issue needs root cause analysis.

## What Claude Should Optimize For
- Narrows the failing path
- Avoids guessy fixes
- Explains root cause clearly

## Workflow
1. Collect the exact failure signal.
2. Map recent changes and affected paths.
3. Test the smallest likely hypotheses.
4. Patch only the verified cause.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- debugging
- root cause
- logs
- tests

---
name: test-generator
description: Help Claude add focused unit, integration, or regression tests that prove behavior without brittle assertions.
---

# Test Generator

## When To Use
Use after adding or changing behavior, especially around edge cases and bug fixes.

## What Claude Should Optimize For
- Covers important paths
- Keeps tests readable
- Avoids over-mocking

## Workflow
1. Identify observable behavior.
2. List nominal and edge cases.
3. Use existing test patterns.
4. Add assertions tied to real outcomes.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- testing
- unit tests
- integration tests
- regression

---
name: refactor-planner
description: Ask Claude to design a safe refactor plan before changing shared modules or tangled code paths.
---

# Refactor Planner

## When To Use
Use before large rewrites, dependency cleanup, file moves, or abstraction changes.

## What Claude Should Optimize For
- Defines a low-risk sequence
- Preserves public behavior
- Highlights rollback points

## Workflow
1. Map current responsibilities.
2. Find hidden callers and contracts.
3. Split the work into reviewable steps.
4. Name validation checks for each step.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- refactor
- architecture
- technical debt
- planning

---
name: api-contract-designer
description: Use Claude to shape API endpoints, request models, responses, errors, and compatibility constraints.
---

# API Contract Designer

## When To Use
Use when designing REST, GraphQL, gRPC, or internal service contracts.

## What Claude Should Optimize For
- Clarifies request and response shape
- Documents error behavior
- Protects backward compatibility

## Workflow
1. Define the caller and job to be done.
2. Model success and failure responses.
3. Check versioning and migration needs.
4. Write implementation and test notes.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- api design
- contracts
- schemas
- compatibility

---
name: database-migration-guard
description: Have Claude review database changes for deploy safety, rollback risk, locks, and application compatibility.
---

# Database Migration Guard

## When To Use
Use before creating, reviewing, or shipping schema migrations.

## What Claude Should Optimize For
- Flags unsafe mixed changes
- Plans phased rollout
- Checks data backfill needs

## Workflow
1. Read the schema and application usage.
2. Classify additive vs destructive changes.
3. Plan forward and rollback behavior.
4. Recommend validation queries.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- database
- migrations
- deploy safety
- rollback

---
name: frontend-component-extractor
description: Let Claude split large React views into focused components while preserving state flow and UI behavior.
---

# Frontend Component Extractor

## When To Use
Use when a page or component becomes too large, hard to test, or hard to reason about.

## What Claude Should Optimize For
- Keeps components focused
- Reduces prop clutter
- Preserves styling

## Workflow
1. Identify natural UI sections.
2. Separate server state from local UI state.
3. Extract components with narrow props.
4. Run type and visual checks.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- React
- components
- frontend
- refactor

---
name: typescript-type-tightener
description: Use Claude to replace loose types with discriminated unions, generics, type guards, and schema-derived models.
---

# TypeScript Type Tightener

## When To Use
Use when code has any, unsafe casts, duplicated DTOs, or weak runtime boundaries.

## What Claude Should Optimize For
- Removes unsafe casts
- Improves inference
- Protects API boundaries

## Workflow
1. Find weakly typed values.
2. Trace source of truth types.
3. Introduce narrow helpers or guards.
4. Verify callers compile without casts.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- TypeScript
- type safety
- schemas
- DTOs

---
name: performance-profiler
description: Guide Claude to diagnose slow pages, endpoints, jobs, or queries using measurements before optimization.
---

# Performance Profiler

## When To Use
Use when something is slow, expensive, memory-heavy, or timing out.

## What Claude Should Optimize For
- Separates symptoms from causes
- Targets the bottleneck
- Avoids premature tuning

## Workflow
1. Capture baseline timing and volume.
2. Find hot paths and repeated work.
3. Choose the smallest measurable change.
4. Compare before and after.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- performance
- profiling
- latency
- optimization

---
name: security-reviewer
description: Ask Claude to review code for auth gaps, injection risk, secrets exposure, unsafe redirects, and data leaks.
---

# Security Reviewer

## When To Use
Use before merging auth, payment, file upload, webhook, admin, or data export changes.

## What Claude Should Optimize For
- Finds exploitable paths
- Checks trust boundaries
- Recommends practical fixes

## Workflow
1. Map inputs and trust boundaries.
2. Check authorization and validation.
3. Review storage and logging of sensitive data.
4. Rank issues by exploitability.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- security
- auth
- privacy
- webhooks

---
name: dependency-upgrade-guide
description: Use Claude to plan package upgrades, breaking change reviews, lockfile updates, and verification steps.
---

# Dependency Upgrade Guide

## When To Use
Use when upgrading frameworks, libraries, SDKs, or language runtimes.

## What Claude Should Optimize For
- Identifies breaking changes
- Scopes required code edits
- Builds a verification plan

## Workflow
1. Read changelogs and current usage.
2. List breaking changes that apply.
3. Patch usage incrementally.
4. Run focused checks and document risks.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- dependencies
- upgrades
- changelogs
- maintenance

---
name: release-notes-writer
description: Turn commits or PRs into clear release notes for users, support, sales, and engineering.
---

# Release Notes Writer

## When To Use
Use before shipping a feature, hotfix, beta, or public changelog entry.

## What Claude Should Optimize For
- Summarizes user impact
- Calls out migration steps
- Separates fixes from features

## Workflow
1. Review merged changes.
2. Group by audience and impact.
3. Translate implementation into outcomes.
4. Include known limitations and follow-ups.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- release notes
- changelog
- shipping
- communication

---
name: deep-research-brief
description: Have Claude produce a structured research brief with sources, assumptions, tradeoffs, and decision-ready takeaways.
---

# Deep Research Brief

## When To Use
Use when exploring a market, technology, vendor, regulation, or unfamiliar domain.

## What Claude Should Optimize For
- Synthesizes multiple sources
- Highlights uncertainty
- Ends with usable recommendations

## Workflow
1. Define the research question.
2. Collect credible sources.
3. Compare claims and contradictions.
4. Summarize findings with confidence levels.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- research
- brief
- analysis
- sources

---
name: competitive-analysis
description: Use Claude to compare competitors by positioning, pricing, features, user segments, and messaging.
---

# Competitive Analysis

## When To Use
Use before positioning a product, writing landing pages, or planning a roadmap.

## What Claude Should Optimize For
- Maps competitive gaps
- Extracts positioning angles
- Finds differentiation opportunities

## Workflow
1. Define the comparison set.
2. Capture public claims and pricing.
3. Compare strengths and weaknesses.
4. Recommend defensible positioning.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- competitors
- positioning
- pricing
- market

---
name: customer-interview-synthesizer
description: Turn interview notes into themes, jobs to be done, objections, quotes, and product opportunities.
---

# Customer Interview Synthesizer

## When To Use
Use after discovery calls, user interviews, sales calls, or support conversations.

## What Claude Should Optimize For
- Finds repeated pain points
- Preserves useful quotes
- Suggests next experiments

## Workflow
1. Clean and group notes.
2. Extract needs and constraints.
3. Separate facts from interpretation.
4. Write themes with evidence.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- customer research
- JTBD
- interviews
- product

---
name: decision-memo
description: Ask Claude to turn messy options into a concise memo with context, criteria, options, tradeoffs, and recommendation.
---

# Decision Memo

## When To Use
Use when a team needs to choose between tools, strategies, vendors, or implementation paths.

## What Claude Should Optimize For
- Makes criteria explicit
- Compares realistic options
- Documents why the choice was made

## Workflow
1. State the decision and deadline.
2. List constraints and stakeholders.
3. Score options against criteria.
4. Recommend the next action.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- decision
- memo
- tradeoffs
- strategy

---
name: source-quality-auditor
description: Use Claude to evaluate whether sources are credible, current, relevant, and suitable for citation.
---

# Source Quality Auditor

## When To Use
Use when a claim depends on external research or could be misleading if sourced poorly.

## What Claude Should Optimize For
- Ranks source credibility
- Flags outdated claims
- Separates primary from secondary evidence

## Workflow
1. List all sources and claims.
2. Check author, date, evidence, and incentives.
3. Mark weak or unsupported claims.
4. Recommend stronger sources.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- sources
- citations
- fact check
- credibility

---
name: market-sizing-assistant
description: Guide Claude through top-down and bottom-up market sizing with assumptions visible and easy to challenge.
---

# Market Sizing Assistant

## When To Use
Use for TAM, SAM, SOM, growth bets, investor materials, or product strategy.

## What Claude Should Optimize For
- Shows assumptions clearly
- Provides sensitivity ranges
- Avoids fake precision

## Workflow
1. Define the exact market.
2. Choose top-down and bottom-up methods.
3. Estimate inputs with ranges.
4. Summarize confidence and risks.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- market sizing
- TAM
- strategy
- forecasting

---
name: literature-review
description: Have Claude synthesize papers, articles, or long documents into themes, evidence strength, and open questions.
---

# Literature Review

## When To Use
Use when reviewing academic, technical, legal, or policy material.

## What Claude Should Optimize For
- Groups findings by theme
- Compares methods and evidence
- Identifies gaps

## Workflow
1. Define the review scope.
2. Extract claims and methods.
3. Cluster papers by theme.
4. Write synthesis with limitations.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- literature
- papers
- synthesis
- evidence

---
name: trend-radar
description: Use Claude to scan weak signals, adoption patterns, risks, and opportunities around an emerging trend.
---

# Trend Radar

## When To Use
Use when deciding whether a new technology, market, or behavior matters.

## What Claude Should Optimize For
- Separates hype from signal
- Identifies adoption drivers
- Suggests monitoring metrics

## Workflow
1. Define the trend and timeframe.
2. Collect examples and counterexamples.
3. Assess incentives and blockers.
4. Create watchlist and next bets.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- trends
- strategy
- signals
- forecasting

---
name: policy-impact-brief
description: Ask Claude to explain how a rule, regulation, or policy change affects a product, team, or market.
---

# Policy Impact Brief

## When To Use
Use when regulations, platform policies, or compliance requirements change.

## What Claude Should Optimize For
- Summarizes obligations
- Highlights operational impact
- Suggests next checks

## Workflow
1. Identify the policy and jurisdiction.
2. Extract relevant clauses.
3. Map impact to workflows.
4. List questions for counsel or owners.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- policy
- compliance
- regulation
- risk

---
name: user-persona-builder
description: Turn research inputs into practical personas with goals, pain points, triggers, objections, and messaging hooks.
---

# User Persona Builder

## When To Use
Use before product positioning, onboarding, feature prioritization, or marketing work.

## What Claude Should Optimize For
- Creates actionable personas
- Links claims to evidence
- Avoids stereotypes

## Workflow
1. Gather user data and quotes.
2. Cluster needs and constraints.
3. Draft personas with jobs to be done.
4. Add product and messaging implications.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- personas
- users
- positioning
- research

---
name: executive-summary
description: Make Claude compress long material into a crisp executive summary with context, implications, and recommended action.
---

# Executive Summary

## When To Use
Use for reports, meeting notes, analysis docs, or strategy updates.

## What Claude Should Optimize For
- Starts with the point
- Keeps nuance without bloat
- Makes next steps clear

## Workflow
1. Read the source material.
2. Extract decision-relevant facts.
3. Group implications by audience.
4. Write a concise summary and action list.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- summary
- executive
- brief
- communication

---
name: technical-docs-writer
description: Use Claude to create developer docs, setup guides, API explanations, and troubleshooting sections.
---

# Technical Docs Writer

## When To Use
Use when documenting a feature, integration, library, workflow, or internal system.

## What Claude Should Optimize For
- Explains prerequisites
- Provides copyable examples
- Anticipates failure modes

## Workflow
1. Identify the reader and goal.
2. Document the happy path.
3. Add examples and edge cases.
4. Include troubleshooting and verification.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- documentation
- developer experience
- guides
- API

---
name: landing-page-copywriter
description: Ask Claude to write conversion-focused landing page copy with hero, benefits, proof, objections, and CTAs.
---

# Landing Page Copywriter

## When To Use
Use when launching a product, feature, waitlist, tool, or offer.

## What Claude Should Optimize For
- Clarifies value proposition
- Connects benefits to pain points
- Improves CTA clarity

## Workflow
1. Define audience and offer.
2. Write headline and subhead options.
3. Structure sections by objections.
4. Polish for clarity and SEO.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- copywriting
- landing page
- conversion
- marketing

---
name: email-sequence-writer
description: Use Claude to create onboarding, sales, lifecycle, or reactivation email sequences with clear goals.
---

# Email Sequence Writer

## When To Use
Use when planning multi-step email communication for users, leads, or customers.

## What Claude Should Optimize For
- Maps each email to intent
- Writes subject lines
- Keeps calls to action focused

## Workflow
1. Define recipient and desired behavior.
2. Choose sequence length and timing.
3. Draft each email with one job.
4. Add subject lines and variants.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- email
- lifecycle
- sales
- onboarding

---
name: tone-rewriter
description: Let Claude rewrite messages in a chosen tone while preserving facts, constraints, and intent.
---

# Tone Rewriter

## When To Use
Use when adapting text for executives, customers, teammates, investors, or public audiences.

## What Claude Should Optimize For
- Preserves meaning
- Matches audience expectations
- Removes awkward phrasing

## Workflow
1. Identify target audience and tone.
2. Mark facts that must not change.
3. Rewrite with clarity.
4. Explain material wording changes.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- tone
- rewriting
- communication
- editing

---
name: proposal-writer
description: Ask Claude to turn a rough idea into a persuasive proposal with problem, solution, plan, risks, and ask.
---

# Proposal Writer

## When To Use
Use for internal initiatives, client work, partnerships, grants, or budget requests.

## What Claude Should Optimize For
- Defines the ask
- Connects benefits to business impact
- Addresses risks

## Workflow
1. Clarify objective and audience.
2. Write problem and proposed solution.
3. Add timeline and resource needs.
4. Close with decision request.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- proposal
- business writing
- strategy
- stakeholders

---
name: social-post-pack
description: Use Claude to create platform-specific social posts, hooks, threads, and repost variations.
---

# Social Post Pack

## When To Use
Use when announcing launches, articles, learnings, case studies, or product updates.

## What Claude Should Optimize For
- Generates multiple angles
- Adapts by platform
- Keeps posts concise

## Workflow
1. Define message and audience.
2. Choose platforms and constraints.
3. Draft hooks and variants.
4. Add CTA and repurposing ideas.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- social media
- content
- launch
- distribution

---
name: case-study-builder
description: Have Claude transform customer wins into credible case studies with problem, solution, metrics, and quotes.
---

# Case Study Builder

## When To Use
Use after a customer success story, pilot, migration, or measurable product impact.

## What Claude Should Optimize For
- Structures the narrative
- Highlights measurable results
- Keeps claims credible

## Workflow
1. Gather customer context and evidence.
2. Map before and after.
3. Write narrative sections.
4. Add quote prompts and proof points.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- case study
- customer story
- proof
- marketing

---
name: weekly-planning-copilot
description: Use Claude to turn goals, projects, and constraints into a realistic weekly plan with priorities.
---

# Weekly Planning Copilot

## When To Use
Use at the start of a week or when too many tasks compete for attention.

## What Claude Should Optimize For
- Ranks priorities
- Protects focus blocks
- Identifies tradeoffs

## Workflow
1. List goals, commitments, and deadlines.
2. Separate must-do from nice-to-do.
3. Schedule focus and admin blocks.
4. Define success for the week.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- planning
- priorities
- weekly review
- focus

---
name: meeting-prep-brief
description: Ask Claude to prepare agendas, context, questions, risks, and desired outcomes before important meetings.
---

# Meeting Prep Brief

## When To Use
Use before customer calls, 1:1s, planning sessions, interviews, or executive reviews.

## What Claude Should Optimize For
- Clarifies meeting purpose
- Prepares smart questions
- Defines decisions needed

## Workflow
1. Provide attendees and context.
2. State desired outcome.
3. Draft agenda and questions.
4. Prepare follow-up template.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- meetings
- agenda
- prep
- questions

---
name: meeting-notes-action-extractor
description: Turn raw notes or transcripts into decisions, action items, owners, deadlines, and unresolved questions.
---

# Meeting Notes Action Extractor

## When To Use
Use after meetings, calls, interviews, retrospectives, or planning sessions.

## What Claude Should Optimize For
- Extracts action items
- Captures decisions
- Flags open questions

## Workflow
1. Clean notes without changing meaning.
2. Identify decisions and owners.
3. Group actions by person or project.
4. Draft follow-up message.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- meeting notes
- actions
- follow-up
- transcripts

---
name: okr-drafter
description: Use Claude to turn strategy and goals into measurable objectives and key results.
---

# OKR Drafter

## When To Use
Use during quarterly planning, team goal setting, or initiative kickoff.

## What Claude Should Optimize For
- Creates measurable KRs
- Avoids output-only goals
- Aligns goals to strategy

## Workflow
1. Capture strategic intent.
2. Draft objectives in plain language.
3. Define measurable key results.
4. Check ambition and ownership.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- OKR
- goals
- planning
- metrics

---
name: prioritization-matrix
description: Ask Claude to rank tasks, bugs, features, or bets by impact, effort, risk, and confidence.
---

# Prioritization Matrix

## When To Use
Use when a backlog or decision list needs ordering.

## What Claude Should Optimize For
- Makes criteria explicit
- Explains rankings
- Surfaces quick wins

## Workflow
1. List candidate items.
2. Choose scoring criteria.
3. Score with assumptions visible.
4. Recommend sequence and deferrals.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- prioritization
- backlog
- impact
- effort

---
name: personal-knowledge-librarian
description: Use Claude to organize notes into summaries, tags, links, reusable snippets, and future retrieval paths.
---

# Personal Knowledge Librarian

## When To Use
Use when notes, articles, docs, or ideas have piled up and need structure.

## What Claude Should Optimize For
- Creates a useful taxonomy
- Extracts reusable notes
- Improves retrieval

## Workflow
1. Collect raw notes.
2. Cluster by theme and project.
3. Write concise summaries.
4. Create tags and next actions.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- notes
- knowledge management
- summaries
- organization

---
name: habit-design-coach
description: Have Claude design practical habits with triggers, friction reduction, tracking, and recovery plans.
---

# Habit Design Coach

## When To Use
Use when building routines for work, health, learning, or creative output.

## What Claude Should Optimize For
- Starts small
- Plans for failure
- Makes tracking simple

## Workflow
1. Define the desired identity and outcome.
2. Choose a tiny behavior.
3. Attach it to an existing cue.
4. Create fallback and review rules.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- habits
- behavior design
- routine
- tracking

---
name: personal-crm-assistant
description: Use Claude to plan relationship follow-ups, context notes, warm intros, and lightweight networking systems.
---

# Personal CRM Assistant

## When To Use
Use after conferences, sales calls, recruiting conversations, or community work.

## What Claude Should Optimize For
- Creates follow-up messages
- Tracks relationship context
- Suggests next touchpoints

## Workflow
1. List contacts and context.
2. Group by relationship goal.
3. Draft personalized follow-ups.
4. Create reminders and next actions.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- networking
- CRM
- follow-up
- relationships

---
name: sql-query-helper
description: Ask Claude to design, explain, optimize, or debug SQL queries with schema context and expected output.
---

# SQL Query Helper

## When To Use
Use when writing analytics, reporting, investigation, or data quality queries.

## What Claude Should Optimize For
- Produces readable SQL
- Explains joins and filters
- Flags performance concerns

## Workflow
1. Provide schema and desired result.
2. Draft the query with clear CTEs.
3. Check edge cases and duplicates.
4. Suggest indexes or validation queries.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- SQL
- analytics
- queries
- debugging

---
name: dashboard-spec-writer
description: Use Claude to define dashboard audiences, metrics, filters, visualizations, and alert thresholds.
---

# Dashboard Spec Writer

## When To Use
Use before building executive, product, operations, finance, or support dashboards.

## What Claude Should Optimize For
- Clarifies metric definitions
- Prevents chart clutter
- Maps questions to visuals

## Workflow
1. Identify the audience and decisions.
2. Define metrics and dimensions.
3. Choose visualizations.
4. Add refresh, ownership, and alert notes.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- dashboards
- metrics
- BI
- visualization

---
name: metric-definition-auditor
description: Have Claude audit metric definitions for ambiguity, double counting, time windows, and ownership.
---

# Metric Definition Auditor

## When To Use
Use when teams disagree about numbers or dashboards show inconsistent results.

## What Claude Should Optimize For
- Clarifies definitions
- Finds data quality gaps
- Documents caveats

## Workflow
1. List current definitions and sources.
2. Check filters and time windows.
3. Find duplicate or missing records.
4. Propose canonical definitions.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- metrics
- data quality
- analytics
- governance

---
name: incident-postmortem
description: Turn timelines, logs, and notes into a blameless postmortem with root cause and preventive actions.
---

# Incident Postmortem

## When To Use
Use after outages, degraded service, data incidents, or operational failures.

## What Claude Should Optimize For
- Builds a timeline
- Separates triggers from root cause
- Creates preventive action items

## Workflow
1. Collect facts and timeline.
2. Identify customer and business impact.
3. Analyze contributing factors.
4. Write corrective actions with owners.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- incident
- postmortem
- operations
- reliability

---
name: runbook-generator
description: Use Claude to write operational runbooks with symptoms, checks, commands, escalation, and rollback steps.
---

# Runbook Generator

## When To Use
Use before handing off recurring operational tasks or on-call procedures.

## What Claude Should Optimize For
- Creates step-by-step checks
- Defines escalation paths
- Includes rollback criteria

## Workflow
1. Describe the system and failure modes.
2. List diagnostic signals.
3. Write commands and expected results.
4. Add escalation and cleanup steps.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- runbook
- on-call
- operations
- reliability

---
name: experiment-analysis
description: Ask Claude to help interpret A/B tests, product experiments, and before-after analyses without overclaiming.
---

# Experiment Analysis

## When To Use
Use when reviewing experiment setup, results, significance, or product implications.

## What Claude Should Optimize For
- Checks validity
- Explains uncertainty
- Recommends next experiments

## Workflow
1. Define hypothesis and metrics.
2. Review sample and assignment quality.
3. Analyze effect size and confidence.
4. Translate results into action.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- experiments
- A/B testing
- statistics
- product

---
name: socratic-tutor
description: Use Claude as a tutor that teaches through questions, hints, examples, and checks for understanding.
---

# Socratic Tutor

## When To Use
Use when learning a topic deeply instead of just getting an answer.

## What Claude Should Optimize For
- Builds understanding gradually
- Adapts to mistakes
- Checks retention

## Workflow
1. Ask for the learner level and goal.
2. Explain the concept briefly.
3. Ask guided questions.
4. Give feedback and next challenge.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- tutoring
- learning
- Socratic
- practice

---
name: interview-practice-coach
description: Have Claude simulate interviews, ask follow-ups, score answers, and suggest sharper responses.
---

# Interview Practice Coach

## When To Use
Use for technical, behavioral, product, sales, or leadership interview prep.

## What Claude Should Optimize For
- Runs realistic practice
- Gives structured feedback
- Improves answer clarity

## Workflow
1. Define role and interview style.
2. Ask one question at a time.
3. Evaluate with a rubric.
4. Rewrite answers using stronger structure.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- interviews
- career
- practice
- feedback

---
name: study-plan-builder
description: Use Claude to build a learning plan with milestones, exercises, projects, review loops, and accountability.
---

# Study Plan Builder

## When To Use
Use when learning a new skill, tool, domain, language, or certification topic.

## What Claude Should Optimize For
- Creates a realistic path
- Balances theory and practice
- Adds review checkpoints

## Workflow
1. Assess current level and deadline.
2. Break the topic into modules.
3. Assign exercises and projects.
4. Schedule review and assessment.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- study plan
- learning
- curriculum
- practice

---
name: concept-explainer
description: Ask Claude to explain hard concepts using analogies, examples, counterexamples, and progressive depth.
---

# Concept Explainer

## When To Use
Use when a topic feels confusing, abstract, or overloaded with jargon.

## What Claude Should Optimize For
- Starts simple
- Adds depth gradually
- Checks misconceptions

## Workflow
1. State the concept and learner background.
2. Explain in plain language.
3. Add examples and non-examples.
4. Ask comprehension checks.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- explanation
- analogies
- education
- clarity

---
name: flashcard-maker
description: Turn notes, docs, or transcripts into spaced-repetition flashcards with precise questions and answers.
---

# Flashcard Maker

## When To Use
Use when preparing for exams, certifications, onboarding, or memorization-heavy topics.

## What Claude Should Optimize For
- Creates atomic cards
- Avoids vague prompts
- Separates recall from recognition

## Workflow
1. Extract key facts and concepts.
2. Create one idea per card.
3. Include examples where useful.
4. Group cards by topic and difficulty.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- flashcards
- spaced repetition
- memory
- study

---
name: deliberate-practice-drills
description: Use Claude to design targeted drills that isolate weak skills and give immediate feedback.
---

# Deliberate Practice Drills

## When To Use
Use when improvement requires repeated practice, not more reading.

## What Claude Should Optimize For
- Targets specific weaknesses
- Creates graded exercises
- Provides feedback loops

## Workflow
1. Identify the skill and weak subskill.
2. Design short drills.
3. Set scoring criteria.
4. Review performance and increase difficulty.

## Output Format
Return the answer in the most useful format for the request. Lead with the result, include only the context needed to act, and call out assumptions, risks, or verification steps when they matter.

## Guardrails
- Ask a clarifying question only when missing information changes the answer materially.
- Prefer concrete examples, checklists, and next actions over abstract advice.
- Keep the response concise unless the user asks for depth.
- Do not invent facts, sources, metrics, or code behavior.

## Tags
- practice
- drills
- feedback
- skill building