Browse skills

Generates conventional commit messages from diffs with scope, breaking change flags, and issue refs.

Four hard rules that stop Claude from silently over-engineering, making wrong assumptions, or touching code it was never supposed to.

Generate or update a repository CHANGELOG.md in Keep a Changelog format from git history - grouping commits since the last tag into Added/Changed/Deprecated/Removed/Fixed/Security, rewriting them as user-facing entries, and recommending the semantic version bump. Use when someone asks "generate a changelog from the git log", "update CHANGELOG.md for this release", "what version bump does this release need", or "turn these commits into release notes". Do NOT use for writing benefit-first product announcement changelogs for end users from release notes - use changelog-writer instead; this skill produces the versioned CHANGELOG.md file that lives in the repo.

Write and review TypeScript as if strict mode plus the hardening flags (noUncheckedIndexedAccess, exactOptionalPropertyTypes) are enforced, and migrate loose codebases to full strictness flag by flag without breaking the build. Use when someone asks "how do I get rid of any", "turn on strict mode in an existing project", "is this type assertion safe", "why does tsc say possibly undefined", or when writing new TypeScript that must pass a strict compiler. Do NOT use for migrating between language or runtime versions - use language-version-migrator instead.

Run a systematic multi-pass code review - correctness, design, security, performance, tests - and report findings ordered by severity with concrete, respectful suggestions. Use when someone asks "review this PR", "review this diff", "what's wrong with this change", or wants a pre-merge quality gate on a branch. Do NOT use for a security-only deep audit of a change - use secure-code-review instead - or for writing the ticket or tracking artifact that describes the change - use jira-ticket-writer instead.

Generates a complete, professional README from a codebase scan - badges, usage, API docs, contributing guide.

Drive development with strict Red-Green-Refactor discipline - write one failing test, write the minimum code to pass it, refactor on green - including test-list planning and a worked kata cycle. Use when someone asks "write this test-first", "do TDD on this feature", "help me practice red-green-refactor", "write a failing test for this bug", or wants tests to drive a new module's design. Do NOT use for adding tests to existing untested legacy code before a refactor - use characterization-test-writer instead - or for designing shared fixtures and factories - use test-data-builder instead.

Turns a git diff into a structured pull request description with motivation, changes, and test plan.

Turns a natural-language video brief into a complete, ready-to-preview Remotion composition - extracts duration, scenes, brand colors, aspect ratio, and real copy; plans the frame budget; and writes data-driven React/TypeScript using useCurrentFrame, interpolate, spring, AbsoluteFill, and Sequence, registered in Root.tsx. Use when someone says "build a 20-second product demo video", "animate a feature walkthrough", "write the Remotion code for this marketing clip", or wants a scene-by-scene composition they can scrub in Studio. Do NOT use for rendering, iterating, or batching the finished MP4 - use remotion-render instead - or for installing and scaffolding the project - use remotion-setup instead.

Scaffolds a new Remotion video project wired for Claude Code Agent Skills - Node check, create-video scaffold, skills install, folder conventions, Google Fonts, and a smoke-test render. Use whenever someone wants to start making videos with Remotion and Claude, even just "make product videos with Claude."

Writes a brand guidelines document - personality traits with is/is-not definitions, voice principles, a tone-by-context matrix, writing mechanics, a say-this-not-that word list, and visual usage rules - all made concrete with do/don't examples. Use when someone asks "write our brand guidelines", "document our voice and tone", "make a style guide the whole team can follow", or teams keep producing off-brand copy and design. Do NOT use for inventing the brand name itself - use brand-naming instead; for organizing messaging by audience and claim, use messaging-hierarchy.

Write LinkedIn posts built for the feed - a hook that survives the two-line truncation, scannable one-idea-per-line structure, a comment-bait close, and first-hour engagement moves. Use when someone asks "write a LinkedIn post about this", "punch up my LinkedIn hook", "why did my post get no reach", or "turn this win into a LinkedIn post". Do NOT use for Twitter/X threads - use tweet-thread-builder instead. Do NOT use for scheduling across a monthly calendar - use social-content-calendar instead.

Write landing-page copy section by section - hero, social proof, features-as-benefits, objections, CTA - leading with one promise and proving it down the page. Use when someone asks "write my landing page", "improve my hero headline", "what should the homepage say", "turn these features into benefits", or is shipping a new page for a launch or campaign. Do NOT use for testing and optimizing an existing page's conversion rate - use landing-page-cro instead; for the one-line claim itself use positioning-statement; for the full message ladder use messaging-hierarchy.

Write recurring email newsletters - subject lines that earn the open, preview text that extends them, a body structure readers finish, and cadence rules that keep the list healthy. Use when someone asks "write my newsletter", "give me subject line options", "why is my open rate dropping", "should I send weekly or monthly", or "help me structure this issue". Do NOT use for automated onboarding or behavior-triggered sequences - use email-drip-builder instead. Do NOT use for inbox placement and spam-folder problems - use email-deliverability instead.

Audit content for publish readiness with the 80-item CORE-EEAT benchmark - per-item Pass/Partial/Fail, dimension and GEO/SEO system scores, critical-trust veto checks, and a prioritized fix plan.

Turn one idea into a tight tweet thread - a hook that earns the click, one beat per post, and momentum to a single CTA. Use when someone asks "turn this into a thread", "write a thread about my launch", "why did my thread flop", or "give me hook options for this". Do NOT use for LinkedIn-format posts - use linkedin-post-writer instead. Do NOT use for adapting one finished piece across many platforms - use content-repurposing instead.

Write one platform-native caption from a topic and brand voice, with a front-loaded hook, native length, and a single earned CTA. Use when the user gives a topic (and optionally brand voice or a described image/video) and asks for an Instagram, LinkedIn, X/Twitter, or TikTok caption. Do NOT use when the user wants a dedicated standalone LinkedIn post - use linkedin-post-writer instead; or a multi-post X thread - use tweet-thread-builder instead; or to adapt one existing piece of content into versions for several channels - use cross-platform-reformatter instead.

Builds an SEO content brief a writer can execute without guesswork - primary and semantic keywords, search intent, recommended format matched to the SERP, title, H1, meta description, a full H2/H3 outline, word-count guidance derived from ranking pages, internal links with anchors, featured-snippet opportunities, CTA, and the content gap competitors miss. Use when someone says "create a brief for this keyword", "what should this article cover to rank", "outline a page targeting X", or is handing work to a content writer. Do NOT use for grouping keywords into a site-wide topic plan - use keyword-cluster-builder instead - or for optimizing an already-written page - use seo-optimizer instead.

Strip predictable AI tells from prose - filler openers, adverbs, binary contrasts, em dashes, vague declaratives, and pull-quotes - with a checklist and a 50-point scoring rubric for revision.

Write short-form video scripts for TikTok, Reels, and Shorts promoting affiliate products with hooks and CTAs.

Runs a thorough multi-source research process - decompose the question, gather independent sources, cross-check, synthesize - and delivers a structured, cited brief with confidence levels and gaps. Use when someone asks "research this topic in depth", "give me everything on X", "what does the evidence say about Y", or needs a broad question answered from secondary sources with citations. Do NOT use for designing a study that collects new data - use primary-research instead; for distilling an already-gathered pile of 20+ sources into themes, use research-synthesis; for verifying one discrete claim, use fact-checker.

Scan social platforms for top-performing content by engagement before creating.

Research topics by collecting real source articles and generating structured research briefs for content creation.

Map competitors across positioning, product, pricing, proof, and weaknesses, then turn the research into sales-ready battlecards with an update cadence. Use when someone asks "what are our competitors doing", "build a battlecard", "how do we compare to X", "why are we losing deals to X", or before a pricing or positioning decision. Do NOT use for designing long-term defensibility strategy - use competitive-moat instead; for a public comparison landing page use comparison-page-builder; for rep-facing objection scripts beyond the battlecard use objection-handler.

Analyze website traffic, global rank, engagement metrics, and traffic sources for domains.

Analyze competitor affiliate strategies, programs, and content to identify profitable niches.

Conducts comprehensive security audits including threat modeling, vulnerability assessment, and remediation guidance.

Find intersection niches where you're the only voice using domain crossing strategy.

Produce evidence-rich research memos and decision briefs with cited sources, verified claims, and explicit limitations - single-agent-first, fanning out only when parallel work clearly helps.

Research and evaluate affiliate programs to find the best ones to promote based on commission rates and niche fit.

Adapt existing content into multiple platform-specific formats with affiliate links and FTC compliance.

Teaches Claude to use the Skill Me catalog itself - browse, install, and manage skills from any conversation. The first skill on the shelf.

Transforms one blog post into 15-30 platform-native micro-content pieces optimized for each social platform.

Manage a three-tier (HOT/WARM/COLD) memory lifecycle for long-running SEO/GEO programs - capture, promote, demote, and archive findings, track open loops, and run privacy cleanup.

Read data from a Google Sheet and create a formatted Google Docs report.

Writes ready-to-build Jira stories with user-story summaries, Given/When/Then acceptance criteria, subtasks, and story-point guidance, checked against INVEST and a Definition of Ready. Use when someone asks "write a Jira ticket for this", "turn these meeting notes into a story", "are these acceptance criteria testable", "should I split this story", or "how many points is this". Do NOT use for configuring Linear teams, cycles, labels, or triage workflows - use linear-workflow instead. For deciding what fits in the sprint, use sprint-planning.

Export Google Contacts directory to a Google Sheets spreadsheet.

Monitor blog content for ranking drops and generate prioritized refresh queue.

Design and optimize post-signup user activation flows to accelerate time-to-value and reduce early churn.

Google Calendar: Manage calendars and events.

Build multi-metric SEO/GEO stakeholder reports - executive summary, organic traffic, rankings, GEO/AI citations, authority, and content quality - each metric source-tagged and compared to a prior period.

Fast structured generation and serving for LLMs with RadixAttention prefix caching. Use for JSON/regex outputs, constrained decoding, agentic workflows with tool calls, or when you need 5× faster inference than vLLM w…

Design and automate surveys, benchmarks, and data collection to create unique competitive advantages through original research.

Parameter-efficient fine-tuning for LLMs using LoRA, QLoRA, and 25+ methods. Use when fine-tuning large models (7B-70B) with limited GPU memory, when you need to train <1% of parameters with minimal accuracy loss, or …

Produces a rigorous, honest evaluation report for an ML model - real baselines, business-matched metrics with confidence intervals, slice-level error analysis, calibration checks, and a go/no-go recommendation. Use when someone asks "is this model good enough to ship", "write up the eval for this model", "did we actually beat the old model", or before any model promotion, stakeholder demo, or retraining decision. Do NOT use for evaluating LLM or prompt outputs - use llm-evaluation instead; for analyzing randomized product experiments use ab-test-analyzer; for documenting a shipped model for consumers use model-card-writer.

Apply prioritized PostgreSQL query, connection, schema, concurrency, RLS, access, and monitoring practices.

Framework for state-of-the-art sentence, text, and image embeddings. Provides 5000+ pre-trained models for semantic similarity, clustering, and retrieval. Supports multilingual, domain-specific, and multimodal models.…

Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1/M2/M3 Macs, AMD/Intel GPUs, or when CUDA is unavailable. Supports GGUF quantization (1.5-8 bit) for redu…

Evaluates LLMs across 100+ benchmarks from 18+ harnesses (MMLU, HumanEval, GSM8K, safety, VLM) with multi-backend execution. Use when needing scalable evaluation on local Docker, Slurm HPC, or cloud platforms. NVIDIA'…

Filesystem RAG benchmarks: corpus/, train.json, evaluate_rag.py (RAGAS quality). Not for prod monitoring, latency/throughput benchmarking (use rag-perf), or evals outside this repo layout.

Builds an accessible brand color palette - one anchor color, role assignments, 9-step tonal scales, 60-30-10 distribution, and documented usage rules - ready for a developer to implement without guessing. Use when someone asks "pick our brand colors", "build a color system for our app", "what shade should our buttons be", or is defining the color layer of a design system. Do NOT use for auditing or fixing contrast failures and color-blindness issues in an existing product - use color-accessibility instead; for assembling the full brand book around the palette, use brand-guidelines.

Chooses and pairs typefaces and builds a modular type scale - ratio, named roles, line heights, measure limits, and responsive values - as a complete design-system type layer. Use when someone asks "what fonts should we use", "build our type scale", "why does our typography feel inconsistent", or is establishing the typographic voice for a brand or product. Do NOT use for animated or motion typography - use kinetic-typography instead; for page-level emphasis, spacing, and layout decisions, use visual-hierarchy; for assembling the full brand book, use brand-guidelines.

Generates twenty-plus brand or product name candidates across five naming archetypes, scores them against weighted criteria without averaging, and runs linguistic and memorability safety checks. Use when someone asks "help me name my startup", "is this a good product name", "generate name ideas for this feature", or is auditing an existing name for distinctiveness, strategic fit, and risk. Not a substitute for professional trademark clearance. Do NOT use for crafting the positioning the name must express - use positioning-statement instead; for briefing the visual identity around a chosen name, use logo-brief-writer.

50+ visual styles, 161 color palettes, 57 font pairings across React, Next.js, SwiftUI, React Native, and more.

Diagnoses and fixes visual hierarchy in a page, screen, or layout using ranked levers - size, weight, color, position, whitespace - and the one-focal-point-per-view rule. Use when someone asks "why does this layout feel flat", "what should the eye see first", "everything is competing on this screen", or wants the reading order of a composition made intentional. Do NOT use for a full multi-dimension design review covering usability and craft - use design-critique instead; for building the underlying type scale, use typography-system.

Designs or audits an icon set so every glyph shares one grid, stroke weight, corner language, naming convention, and export pipeline, and delivers the written icon spec engineering builds against. Use when someone asks "our icons look inconsistent", "set up an icon grid", "how should we name icons", "audit this icon set", "prep our SVGs for handoff", or is commissioning new icons and needs the rules first. Do NOT use for a single logo mark - use logo-brief-writer instead; for delivering full design specs beyond icons, use design-handoff-doc.

Writes a complete creative brief for a logo or identity project - business context, a three-adjective personality axis, deliverables and constraints, competitive territory, rounds and timeline, and concrete success criteria. Use when someone asks "write a logo brief", "brief a designer for our rebrand", "what do I send the identity studio", or is kicking off a brand identity engagement. Do NOT use for generating the brand name itself - use brand-naming instead; for documenting the finished identity system's usage rules, use brand-guidelines; for briefing broader creative direction, use moodboard-builder.

Assembles a curated moodboard of 8 to 16 images plus a 150-300 word art direction statement that argues for one visual direction. Use when someone asks "build a moodboard", "set the visual direction for this brand", "get the team aligned on look and feel before design starts", or is presenting art direction to a client. Do NOT use for sourcing and licensing individual production images - use visual-asset-curation or stock-photo-finder instead; for codifying an already-approved identity into usage rules, use brand-guidelines.

Official Anthropic skill. Guides Claude to create distinctive, production-grade UIs - not the default Inter + purple gradient.

Designs marketing-site hero and launch-page motion - scroll-linked scrub sections pinned 100-200vh, one signature mechanic per page, small masked reveals, and hard performance gates - so the page choreographs the story without scroll-jacking, jank, or the animated-template look. Use when someone asks "add a scroll animation to our landing page hero", "pin a section that scrubs as the user scrolls", "is this scroll-jacking or scroll-linked", "our homepage animations feel cheap and janky", or "what should move on our launch page", or when a launch page needs motion that still passes performance budgets. Do NOT use for in-product UI motion systems - use animation-system instead; for the hero's static artwork and composition, use website-hero-artwork; for the headline words themselves, use landing-page-copy; for a rendered video file embedded in the hero, use remotion-compose.

Designs a SaaS pricing model - the value metric, three-tier packaging with feature gates, list prices, annual discount, and a built-in expansion path targeting net revenue retention - delivered as a pricing page spec. Use when a founder asks "how should we price our SaaS", "per seat or usage-based", "what features go in each tier", or is building or overhauling a pricing page. Do NOT use for pricing strategy outside SaaS packaging - use pricing-strategy instead; for the upsell motion after pricing exists, use expansion-revenue.

Use when turning a positioning statement into actual copy - building the value proposition, message pillars, proof points, and per-channel messaging that keep the website, ads, and sales deck all saying the same thing. Triggers on "write our value proposition", "message pillars", "messaging framework", "messaging house", "key messages", "proof points", "our website and sales say different things", "per-channel messaging", "on-message copy", "get everyone on the same message". Takes the positioning-statement as input and feeds landing-page-copy. Do NOT use when you have not yet fixed positioning (audience, category, differentiator) - use positioning-statement first. Do NOT use when you need the actual hero/landing page words - use landing-page-copy with this framework as the brief. Do NOT use to schedule or sequence the launch phases (pre-launch, launch, post-launch) - use launch-plan-sequencer. Do NOT use to plan the hour-by-hour go-live - use launch-day-runbook. Do NOT use to build the sales deck, one-pager, or talk track - use sales-enablement-kit (it consumes this house). Do NOT use to design the in-product / free-trial activation flow - use plg-motion-designer (it pulls its language from this house).

Audit technical SEO health - crawlability, indexing, Core Web Vitals, robots.txt, sitemaps, canonicals, redirects, HTTPS, and structured data - with scored results and a prioritized fix roadmap.

Build a competitor-relative coverage map - missing topics, keyword gaps, and format gaps - bucketed into Quick Wins, Strategic Builds, and Long-term, with a dated content-calendar entry per quick win.

Build a sequenced go-to-market plan - beachhead ICP, positioning, one or two channels, and dated milestones - instead of a scattershot launch. Use when someone asks "how do I take this to market", "what's my GTM strategy", "which channel should I start with", "who should I sell to first", or is planning a new product or a new market entry. Do NOT use for the dated launch-week timeline and asset checklist - use launch-plan-sequencer instead; for Product Hunt copy and day-of tactics use product-hunt-launch; for deep persona research use icp-persona-builder.

Analyze a backlink profile - referring domains, anchor-text mix, toxic-link share, profile health, and competitor link gaps - then recommend concrete link-building or disavow actions.

Write and improve title tags, meta descriptions, Open Graph, and Twitter Cards for CTR and sharing - three on-spec options each within character limits, plus a complete social/canonical tag block.

Reposition a generalist agency around a vertical or service specialization, write the "we're the agency that X for Y" statement, and design a paid-audit entry offer backed by aligned case studies. Use when an agency owner says "we do everything for everyone and it's killing our pricing", "should we niche down", "prospects can't tell us apart from every other agency", or "how do we productize the front of our funnel". Do NOT use for writing a one-line positioning statement for a product or startup - use positioning-statement instead.

Estimate market size and validate market opportunities using TAM/SAM/SOM frameworks.

Use when planning a product launch end-to-end and you need the full dated timeline. Triggers on "plan my launch", "launch plan", "how do I sequence a launch", "pre-launch checklist", "what happens before launch day", "launch timeline", "GTM launch plan", "coordinate a launch across channels", "who owns what for launch". Builds the pre-launch → launch-day → post-launch arc with a channel & asset checklist, owners, and dates. Do NOT use when you need the minute-by-minute run-of-show for the day itself - use launch-day-runbook instead. Do NOT use to write the core message - use positioning-statement for the one-line claim and messaging-hierarchy for the laddered copy. Do NOT use to design the self-serve adoption motion - use plg-motion-designer. Do NOT use to build the sales deck, battlecard, or demo script - use sales-enablement-kit.

Use when the user has a chosen workout and wants it turned into a coached class or lesson plan (e.g. "turn this into a 60-minute class", "build a class plan for this WOD"). Produces a minute-by-minute timeline with timing, coach positioning, teaching points, scaling checkpoints, transitions, buffers, and a late-running contingency; defaults to 60 minutes. Excludes designing the workout (route to crossfit-workout-designer) and large-class floor logistics/equipment-sharing (crossfit-class-logistics-and-floor-manager owns heats and station maps).

Use when a CrossFit athlete has pain, an injury, a clinician restriction, or is returning to training and needs a session or plan adapted around it. Expects the stated restriction or clinician limits, current symptoms, training goal, and equipment. Modifies training around limits without diagnosing or rehabbing - diagnosis and rehab route to a clinician, generic difficulty scaling to crossfit-stimulus-and-scaling-coach, and mobility limitation without pain to crossfit-mobility-assessment-and-programmer.

Use when the user asks for a single CrossFit / functional-fitness workout, WOD, or metcon ("something for today"), supplying or implying a stimulus, time domain, level, or equipment. Designs one complete workout (stimulus, time domain, patterns, loading, scaling paths) from those inputs plus optional maxes and phase. Excludes multiweek programming, owned by crossfit-programming-cycle-builder; scaling an existing workout, owned by crossfit-stimulus-and-scaling-coach; and class plans, owned by crossfit-class-plan-builder.

Use when an athlete reports a position limitation and wants to improve it ("can't reach overhead", "no squat depth", "can't hit the front rack"). Expects the position, what it limits, and any pain. Distinguishes passive vs. active mobility vs. motor control vs. strength vs. familiarity vs. pain via a discriminating test, then builds a short plan with an objective retest. Excludes pain-led modification (route to crossfit-injury-aware-workout-adapter) and full assessment (defer to crossfit-athlete-assessment); never diagnoses tissue.

Use when the user has a SPECIFIC released or known CrossFit competition workout and wants a strategy to attack it (goal score, rep-break plan, pacing, no-rep risk, repeat decision). Expects the official workout description and standards, supplied or confirmed by the user, plus the athlete's capacities. When standards are missing it returns a CONDITIONAL strategy and requests the text; it never invents standards. Excludes designing a workout, owned by crossfit-workout-designer, and rules, owned by crossfit-rules-and-standards-advisor.

Use when the user wants to develop, learn, or progress a specific CrossFit movement over weeks - phrasings like "help me get my first muscle-up", "build my snatch", "I want handstand walks". Builds a development plan: prerequisites, assessment tests, limiting factors, teaching sequence, drills, accessory work, weekly dosage, advancement criteria, and a retest. Excludes diagnosing a fault in a movement already being performed (crossfit-movement-fault-analyzer owns that) and scaling a single workout (crossfit-stimulus-and-scaling-coach).

Use when a coach wants cues to fix or improve an already-named movement or fault (e.g. "cues for the squat", "cues to fix early arm pull", "how to cue kipping rhythm"). Generates cues by type - verbal, visual, tactile (consensual only), external/internal-focus, constraint, drill - then prioritizes the one or two best and says when to stop talking. Expects the movement plus the fault. Excludes diagnosing faults from raw observation, owned by crossfit-movement-fault-analyzer, and building a skill over weeks, owned by crossfit-movement-progression-builder.

Use when an athlete or coach wants a multi-week PLAN, peaking timeline, or season roadmap leading INTO a CrossFit competition (the Open, a qualifier, a local throwdown, a multi-event, or a team event). Expects target event, date, division, athlete level, and availability. Builds phases (base, competition-specific, peak/taper), standards exposure, and simulations. Excludes attacking a single released workout, owned by crossfit-competition-workout-strategist, and interpreting rules, owned by crossfit-rules-and-standards-advisor.

Use when the user wants an overall evaluation of where a CrossFit athlete stands ("assess me", "build my profile") across strength, power, capacity, gymnastics, weightlifting, mobility, and skill. Expects whatever maxes and benchmarks the athlete supplies; flags unknowns instead of inventing them. Produces a profile of strengths, limiters, unknowns, and recommended tests. Excludes ranking which limiter to fix first once a profile exists (crossfit-weakness-prioritizer) and designing the test battery (crossfit-benchmark-and-testing-builder).

Use when the user lacks a piece of equipment and needs a substitute movement or implement for a workout (e.g. "no rower, what instead?", "no barbell for thrusters"). Maps the movement's demand profile (joint action, force, energy system, skill, impact), proposes the closest feasible swap, and states what it preserves vs. fails to reproduce plus a loading adjustment. Expects the movement and available gear. Excludes general difficulty scaling, owned by crossfit-stimulus-and-scaling-coach, and designing a workout, owned by crossfit-workout-designer.