skill
claude-skills
Collection of Claude Code skills for enhanced AI workflows
Set up
Skills are installed by placing their folder where your agent loads skills.
Install instructions come from the README
Skills do not use an MCP config. Follow the steps in the README below; one-click install arrives with accounts.
README
synced Oct 14, 2026Claude Skills
~100 skills for Claude Code — meeting pipelines, research, image generation, TDD, publishing, personal analytics, and Claude Code ops. Each skill is a self-contained folder with a SKILL.md; Claude triggers them from natural language.
By Gleb Kalinin — Berlin-based technologist, AI educator, solopreneur, artist. Join the Claude Code Lab.
Install
# Plugin marketplace (recommended)
claude plugin marketplace add glebis/claude-skills
claude plugin install tdd@glebis-skills
# skills CLI
npx skills add glebis/claude-skills --skill tdd
# Manual
git clone https://github.com/glebis/claude-skills.git
cp -r claude-skills/<skill-name> ~/.claude/skills/
Per-skill setup (API keys, Python/Node deps) is documented in each skill's SKILL.md. See BUNDLES.md for curated groupings.
Contents
Recent additions · Meetings & transcripts · Research & knowledge · Writing, design & publishing · Images, audio & media · Communication · Development & release · Thinking & decisions · Personal analytics · Claude Code ops · Lab, consulting & business
Recent additions
| Added | Skill | What it does |
|---|---|---|
| 2026-07 | font-features | Inspect and apply OpenType features — ligatures, stylistic sets, alternates |
| 2026-07 | typography | Deterministic smart quotes, dashes, non-breaking spaces (RU/EN/DE/FR) |
| 2026-07 | pre-session-portrait | 7-lens JTBD client portrait before a paid consulting hour |
| 2026-07 | automation-advisor | Quantified ROI + break-even analysis for "should I automate this?" |
| 2026-07 | cull-release | Six-skill release orchestration suite (check, prepare, publish, verify, recover) |
| 2026-07 | elimination-research | Numeric, source-auditable shortlist comparisons for products and vendors |
| 2026-07 | nielsen-heuristics | Formal heuristic evaluation against Nielsen's 10 usability heuristics — moved to the humane plugin |
| 2026-07 | i18n-studio | Edit, translate, and review Astro-style i18n string corpora |
| 2026-07 | the-goal | Theory-of-Constraints diagnostic for deciding what to automate |
| 2026-06 | design-tokens | DTCG 2025.10 token sets — validate, resolve, export to CSS — moved to the humane plugin |
| 2026-06 | app-release | iOS/watchOS release pipeline to TestFlight and the App Store |
| 2026-06 | local-models | Offline llama.cpp inference reusing models Ollama already downloaded |
| 2026-06 | feature-factory | Solo-dev feature pipeline: goal-first, TDD, deterministic verification |
Meetings & transcripts
| Skill | What it does |
|---|---|
| fathom | Fetch Fathom meetings, transcripts, summaries, action items; download video |
| granola | Granola cache/API → transcripts and Obsidian export, with 15-min auto-sync |
| zoom | Create/manage Zoom meetings, access cloud recordings and transcripts |
| meeting-prep | Pull Cal.com bookings, research participants, audit past sessions, write prep notes |
| meeting-processor | Auto-detect meeting type and extract type-specific structured summaries |
| transcript-analyzer | Decisions, action items, opinions, terminology via Cerebras llama-3.3-70b |
| coaching-session-summarizer | Summarize coaching/therapy transcripts after a Fathom/Granola sync |
| session-anonymizer | Three-layer PII anonymization (Natasha NER + privacy filter + local LLM) |
| youtube-transcript | Transcript + metadata → Obsidian markdown, no video download |
| thinking-patterns | Longitudinal cognitive analysis across months of transcripts (12 dimensions, multi-agent) |
Research & knowledge
| Skill | What it does |
|---|---|
| deep-research | OpenAI Deep Research API with prompt enhancement and sourced markdown reports |
| firecrawl-research | Scrape and enrich notes with web sources; scientific writing with citations |
| elimination-research | Deterministic scoring, ownership costs, and audit trail for shortlists |
| doctorg | Evidence-based health research, GRADE-style ratings, Apple Health context |
| qmd-search | Local semantic + keyword search over an Obsidian vault; cross-lingual (EN↔RU) |
| learning-vault | Learning profile for portable study vaults, with optional Dataview |
| open-knowledge-vault | Maintain source-grounded OKF vaults with optional Obsidian and Dataview |
| temple-generator | Turn a vault into a 3D interactive knowledge map with generative audio |
| daydream | Multi-agent mining of non-obvious connections between vault notes |
| insight-extractor | Parse /insights output into trackable markdown action items |
| weekly-digest | Research, score, and publish a weekly industry digest on any topic |
| wow-digest | Daily digest of genuinely surprising items, scored for epistemic friction |
| rag-eval | Tune RAG pipelines with structured evals instead of eyeballing |
| rigorous-experiments | Pre-registered n-of-1 experiments, permutation tests, FDR discipline |
| whitepaper-audit | Audit long-form technical documents against a research-grounded checklist |
| name-audition | Domain availability, collision research, and scoring for candidate names |
| google-image-search | Google CSE image search with LLM-powered selection and Obsidian enrichment |
Writing, design & publishing
| Skill | What it does |
|---|---|
| present | Narrated HTML presentations — dual article/slides mode, ElevenLabs voiceover |
| presentation-generator | Neobrutalist HTML decks with Anime.js; export to PNG/PDF/video |
| pdf-generation | Markdown → PDF via Pandoc; mobile (6×9in) and print (A4) layouts, EN/RU |
| tufte-report | Tufte-style data reports and dashboards as standalone HTML |
| brand-agency | Agency brand colors, typography, and 11 social templates via Playwright |
| design-tokens | DTCG token sets, CSS export, APCA/WCAG contrast — moved to the humane plugin |
| typography | Locale-correct quotes, dashes, and non-breaking spaces via pinned CLI |
| font-features | Inspect and apply ligatures, stylistic sets, and character variants |
| nielsen-heuristics | Evidence-backed heuristic usability evaluation — moved to the humane plugin |
| de-ai | Humanize AI-sounding text across 5 languages, preserving meaning and facts |
| i18n-studio | Edit, translate, and review i18n string corpora |
| sketch | Sketch/diagram generation support |
| publish-skill | Publish or update a skill on the claude-skills-site Astro website |
| telegram-post | Draft → preview → publish formatted Telegram channel posts |
| agency-socials | Event covers, YouTube thumbnails, and social assets for AGENCY Community |
Images, audio & media
| Skill | What it does |
|---|---|
| gpt-image-2 | OpenAI GPT Image 2 — thinking mode, 14 style presets, draft mode, cost controls |
| nano-banana | Gemini image generation and editing with style and platform presets |
| vision-bench | Score and compare images using vision LLMs as judges (11 criteria presets) |
| elevenlabs-tts | Text → audio via ElevenLabs with voice presets and tunable parameters |
Communication
| Skill | What it does |
|---|---|
| telegram | Fetch, search, send, schedule Telegram messages and media |
| telegram-telethon | Full Telethon wrapper + daemon that spawns Claude sessions on triggers |
| tg-responder | Review and send response drafts; track unanswered outbound messages |
| gws | Google Workspace CLI reference — Gmail, Calendar, Drive, Sheets, Docs, Chat |
| gmail | Gmail search, fetch, attachments, labels via API |
| browser-mate | Automate a logged-in Chrome without disturbing the user's open tabs |
Development & release
| Skill | What it does |
|---|---|
| tdd | Multi-agent TDD with enforced context isolation across RED→GREEN→REFACTOR |
| feature-factory | Solo-dev pipeline from feature intent to shipped, with evidence gates |
| skills/release | Config-driven releases with a tiered compatibility policy (COMPATIBILITY.md) |
| cull-release | Cull release suite: check, prepare, publish, verify, recover |
| app-release | iOS/watchOS → TestFlight and App Store |
| init-tauri-app | Scaffold Tauri v2 projects with house conventions |
| init-xcode-app | Scaffold SwiftUI apps via XcodeGen (macOS, iOS, iOS+watchOS) |
| agent-cli | Add agent-friendly --json NDJSON output to Python CLIs |
| repo-prep | Prepare a repo for publication — LICENSE, README, metadata, community docs |
| llm-cli | One CLI across 6 LLM providers and 40+ models |
| local-models | Cheap offline inference via llama.cpp on Ollama-downloaded models |
| codex | Drive OpenAI Codex CLI for analysis, refactoring, automated edits |
| peer-agent-collaboration | Have Claude Code, Codex, and other agents work together as peers |
| linear | Create, list, update, and search Linear issues and projects |
| github-gist | Publish files or notes as secret/public gists |
| skills/job-babysitter | Watch long-running background jobs (encodes, embeddings, batch pipelines) |
| skills/cmux | Control the cmux terminal app via its CLI socket API |
Thinking & decisions
| Skill | What it does |
|---|---|
| the-goal | Theory-of-Constraints diagnostic before building any automation |
| automation-advisor | 8 questions → break-even math, with voice-enabled web UI |
| decision-toolkit | 7 frameworks, 20+ bias counter-questions, interactive HTML wizards |
| jtbd | Jobs-to-be-Done interviews, switching forces, messaging and GTM briefs — moved to the humane plugin |
| balanced | Anti-sycophantic multi-perspective analysis and Socratic dialogue |
| cognitive-toolkit | CBT/DBT interventions — thought records, DEAR MAN, crisis skills (separate repo) |
| synthetic-session-generator | Persona-consistent synthetic coaching/therapy transcripts for evals |
Personal analytics
| Skill | What it does |
|---|---|
| health-data | Query Apple Health SQLite — vitals, activity, sleep, workouts (MD/JSON/FHIR) |
| wispr-analytics | Analyze Wispr Flow dictation history; manage the dictation dictionary |
| browsing-history | Query history across all synced Chrome devices with natural language |
| chrome-history | Query local Chrome history by date, type, keyword, site |
| session-finder | Semantic index of Claude Code sessions (Gemini embeddings); relaunch best match |
| session-search | Search session transcripts with synonym-aware semantic matching |
Claude Code ops
| Skill | What it does |
|---|---|
| ecosystem | Audit skill health, project pulse, CLAUDE.md drift, service status |
| retrospective | Extract learnings from one session or all of today's, update skills and memory |
| skill-studio | Interview-driven skill/automation design; analyze sessions into new skills |
| recording | Demo mode — redact PII from outputs in real time while screen-sharing |
| skills/disk-cleanup | Survey and clean macOS caches, package data, crash dumps |
| wispr-fix | Queue and batch-apply Wispr dictation corrections |
| context-builder | Generate AI-transformation discovery prompts for consulting clients |
| meta | Run Claude Code commands against the telegram_agent project |
Lab, consulting & business
| Skill | What it does |
|---|---|
| agency-docs-updater | Lab meeting pipeline: transcript → video → YouTube → summary → MDX → deploy |
| agency-meetup-publish | Zoom recording → intro/outro, thumbnail, timecodes, YouTube, playlist |
| lab-retro | Four-part retrospective and self-assessment for Lab participants |
| skills/library-sync | Sync bilingual (EN/RU) library content for agency-docs |
| site-diagnosis | Pre-consultation questionnaire for clients building sites with AI tools |
| sorted | getSorted.de automation — invoicing, expenses, German VAT/ZM filings |
| cull | View, rate, organize, search, and export images with the Cull app |
| timebuzzer-led | Control timeBuzzer hardware LED via MIDI — colors, effects, status signals |
Skill structure
skill-name/
├── SKILL.md # Metadata, triggers, usage
├── CHANGELOG.md # Version history
├── scripts/ # Executable orchestration
├── assets/ # Resources
└── references/ # Detailed docs
Contributing
Fork, add a skill following the structure above, document it, open a PR. For guidance on authoring skills, see the Claude Code skills docs.
License
MIT — see individual skill directories for exceptions.
Skills require Claude Code to function; they are not standalone tools. Support this work and learn Claude Code →