Claude Skills

July 29, 2026 · View on GitHub

~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

AddedSkillWhat it does
2026-07font-featuresInspect and apply OpenType features — ligatures, stylistic sets, alternates
2026-07typographyDeterministic smart quotes, dashes, non-breaking spaces (RU/EN/DE/FR)
2026-07pre-session-portrait7-lens JTBD client portrait before a paid consulting hour
2026-07automation-advisorQuantified ROI + break-even analysis for "should I automate this?"
2026-07cull-releaseSix-skill release orchestration suite (check, prepare, publish, verify, recover)
2026-07elimination-researchNumeric, source-auditable shortlist comparisons for products and vendors
2026-07nielsen-heuristicsFormal heuristic evaluation against Nielsen's 10 usability heuristics
2026-07i18n-studioEdit, translate, and review Astro-style i18n string corpora
2026-07the-goalTheory-of-Constraints diagnostic for deciding what to automate
2026-06design-tokensDTCG 2025.10 token sets — validate, resolve, export to CSS
2026-06app-releaseiOS/watchOS release pipeline to TestFlight and the App Store
2026-06local-modelsOffline llama.cpp inference reusing models Ollama already downloaded
2026-06feature-factorySolo-dev feature pipeline: goal-first, TDD, deterministic verification

Meetings & transcripts

SkillWhat it does
fathomFetch Fathom meetings, transcripts, summaries, action items; download video
granolaGranola cache/API → transcripts and Obsidian export, with 15-min auto-sync
zoomCreate/manage Zoom meetings, access cloud recordings and transcripts
meeting-prepPull Cal.com bookings, research participants, audit past sessions, write prep notes
meeting-processorAuto-detect meeting type and extract type-specific structured summaries
transcript-analyzerDecisions, action items, opinions, terminology via Cerebras llama-3.3-70b
coaching-session-summarizerSummarize coaching/therapy transcripts after a Fathom/Granola sync
session-anonymizerThree-layer PII anonymization (Natasha NER + privacy filter + local LLM)
youtube-transcriptTranscript + metadata → Obsidian markdown, no video download
thinking-patternsLongitudinal cognitive analysis across months of transcripts (12 dimensions, multi-agent)

Research & knowledge

SkillWhat it does
deep-researchOpenAI Deep Research API with prompt enhancement and sourced markdown reports
firecrawl-researchScrape and enrich notes with web sources; scientific writing with citations
elimination-researchDeterministic scoring, ownership costs, and audit trail for shortlists
doctorgEvidence-based health research, GRADE-style ratings, Apple Health context
qmd-searchLocal semantic + keyword search over an Obsidian vault; cross-lingual (EN↔RU)
learning-vaultGenerate an Obsidian study vault for any certification or course
temple-generatorTurn a vault into a 3D interactive knowledge map with generative audio
daydreamMulti-agent mining of non-obvious connections between vault notes
insight-extractorParse /insights output into trackable markdown action items
weekly-digestResearch, score, and publish a weekly industry digest on any topic
wow-digestDaily digest of genuinely surprising items, scored for epistemic friction
rag-evalTune RAG pipelines with structured evals instead of eyeballing
rigorous-experimentsPre-registered n-of-1 experiments, permutation tests, FDR discipline
whitepaper-auditAudit long-form technical documents against a research-grounded checklist
name-auditionDomain availability, collision research, and scoring for candidate names
google-image-searchGoogle CSE image search with LLM-powered selection and Obsidian enrichment

Writing, design & publishing

SkillWhat it does
presentNarrated HTML presentations — dual article/slides mode, ElevenLabs voiceover
presentation-generatorNeobrutalist HTML decks with Anime.js; export to PNG/PDF/video
pdf-generationMarkdown → PDF via Pandoc; mobile (6×9in) and print (A4) layouts, EN/RU
tufte-reportTufte-style data reports and dashboards as standalone HTML
brand-agencyAgency brand colors, typography, and 11 social templates via Playwright
design-tokensDTCG token sets — global base + per-project layering, CSS export
typographyLocale-correct quotes, dashes, and non-breaking spaces via pinned CLI
font-featuresInspect and apply ligatures, stylistic sets, and character variants
nielsen-heuristicsEvidence-backed heuristic usability evaluation
de-aiHumanize AI-sounding text across 5 languages, preserving meaning and facts
i18n-studioEdit, translate, and review i18n string corpora
sketchSketch/diagram generation support
publish-skillPublish or update a skill on the claude-skills-site Astro website
telegram-postDraft → preview → publish formatted Telegram channel posts
agency-socialsEvent covers, YouTube thumbnails, and social assets for AGENCY Community

Images, audio & media

SkillWhat it does
gpt-image-2OpenAI GPT Image 2 — thinking mode, 14 style presets, draft mode, cost controls
nano-bananaGemini image generation and editing with style and platform presets
vision-benchScore and compare images using vision LLMs as judges (11 criteria presets)
elevenlabs-ttsText → audio via ElevenLabs with voice presets and tunable parameters

Communication

SkillWhat it does
telegramFetch, search, send, schedule Telegram messages and media
telegram-telethonFull Telethon wrapper + daemon that spawns Claude sessions on triggers
tg-responderReview and send response drafts; track unanswered outbound messages
gwsGoogle Workspace CLI reference — Gmail, Calendar, Drive, Sheets, Docs, Chat
gmailGmail search, fetch, attachments, labels via API
browser-mateAutomate a logged-in Chrome without disturbing the user's open tabs

Development & release

SkillWhat it does
tddMulti-agent TDD with enforced context isolation across RED→GREEN→REFACTOR
feature-factorySolo-dev pipeline from feature intent to shipped, with evidence gates
skills/releaseConfig-driven releases with a tiered compatibility policy (COMPATIBILITY.md)
cull-releaseCull release suite: check, prepare, publish, verify, recover
app-releaseiOS/watchOS → TestFlight and App Store
init-tauri-appScaffold Tauri v2 projects with house conventions
init-xcode-appScaffold SwiftUI apps via XcodeGen (macOS, iOS, iOS+watchOS)
agent-cliAdd agent-friendly --json NDJSON output to Python CLIs
repo-prepPrepare a repo for publication — LICENSE, README, metadata, community docs
llm-cliOne CLI across 6 LLM providers and 40+ models
local-modelsCheap offline inference via llama.cpp on Ollama-downloaded models
codexDrive OpenAI Codex CLI for analysis, refactoring, automated edits
peer-agent-collaborationHave Claude Code, Codex, and other agents work together as peers
linearCreate, list, update, and search Linear issues and projects
github-gistPublish files or notes as secret/public gists
skills/job-babysitterWatch long-running background jobs (encodes, embeddings, batch pipelines)
skills/cmuxControl the cmux terminal app via its CLI socket API

Thinking & decisions

SkillWhat it does
the-goalTheory-of-Constraints diagnostic before building any automation
automation-advisor8 questions → break-even math, with voice-enabled web UI
decision-toolkit7 frameworks, 20+ bias counter-questions, interactive HTML wizards
jtbdJobs-to-be-Done interviews, switching forces, messaging and GTM briefs
balancedAnti-sycophantic multi-perspective analysis and Socratic dialogue
cognitive-toolkitCBT/DBT interventions — thought records, DEAR MAN, crisis skills (separate repo)
synthetic-session-generatorPersona-consistent synthetic coaching/therapy transcripts for evals

Personal analytics

SkillWhat it does
health-dataQuery Apple Health SQLite — vitals, activity, sleep, workouts (MD/JSON/FHIR)
wispr-analyticsAnalyze Wispr Flow dictation history; manage the dictation dictionary
browsing-historyQuery history across all synced Chrome devices with natural language
chrome-historyQuery local Chrome history by date, type, keyword, site
session-finderSemantic index of Claude Code sessions (Gemini embeddings); relaunch best match
session-searchSearch session transcripts with synonym-aware semantic matching

Claude Code ops

SkillWhat it does
ecosystemAudit skill health, project pulse, CLAUDE.md drift, service status
retrospectiveExtract learnings from one session or all of today's, update skills and memory
skill-studioInterview-driven skill/automation design; analyze sessions into new skills
recordingDemo mode — redact PII from outputs in real time while screen-sharing
skills/disk-cleanupSurvey and clean macOS caches, package data, crash dumps
wispr-fixQueue and batch-apply Wispr dictation corrections
context-builderGenerate AI-transformation discovery prompts for consulting clients
metaRun Claude Code commands against the telegram_agent project

Lab, consulting & business

SkillWhat it does
agency-docs-updaterLab meeting pipeline: transcript → video → YouTube → summary → MDX → deploy
agency-meetup-publishZoom recording → intro/outro, thumbnail, timecodes, YouTube, playlist
lab-retroFour-part retrospective and self-assessment for Lab participants
skills/library-syncSync bilingual (EN/RU) library content for agency-docs
site-diagnosisPre-consultation questionnaire for clients building sites with AI tools
sortedgetSorted.de automation — invoicing, expenses, German VAT/ZM filings
cullView, rate, organize, search, and export images with the Cull app
timebuzzer-ledControl 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 →