glebis/claude-skills
81 skills11.7k installs27.5k starsGitHub
Install
npx skills add https://github.com/glebis/claude-skillsSkills in this repo
1Google Image Searchgoogle-image-search is a Claude Code skill from glebis/claude-skills that searches and downloads images via the Google Custom Search API with intelligent scoring and LLM-based selection. The skill supports simple keyword queries, batch processing from JSON config files, automatic config generation from search terms, and full Obsidian note enrichment with images inserted below headings. Developers reach for google-image-search when illustrating technical articles, research documents, or presentations without manually browsing stock sites, and when bulk-enriching knowledge bases stored in Obsidian. Triggers include finding images for articles, adding visuals to presentations, enriching notes, or running batch image downloads from a JSON configuration. The workflow combines API search results with LLM ranking so chosen images match article context while respecting rights-aware selection criteria stated in the skill documentation.606installs2Pdf Generationpdf-generation is a document automation skill from glebis/claude-skills that produces polished PDF reports, invoices, and export packs from markdown, HTML, or structured JSON and CSV data. The skill guides layout, typography, headers, tables, and pagination choices so generated PDFs are suitable for customer delivery rather than raw print-to-PDF dumps. Developers reach for pdf-generation when building SaaS billing exports, monthly analytics reports, compliance document packs, or automated invoice attachments in API backends. Output includes PDF files plus the generating script or template using common Node.js, Python, or headless browser tooling patterns that can be integrated into CI or on-demand API endpoints.437installs3Health Datahealth-data is a glebis/claude-skills agent skill for querying and analyzing Apple Health data stored in a local SQLite database containing 6.3M+ records across 43 health metric types. The health_query.py script supports daily summaries, weekly trends, sleep analysis, vitals, activity rings, workouts, and raw SQL templates. Output formats include Markdown, JSON, FHIR R4 Bundles with LOINC-coded Observations, and ASCII Unicode bar charts. Developers reach for health-data when building health apps, generating fitness reports, exporting interoperable FHIR data, or validating privacy-aware query patterns for clinical and benefits workflows. Reference files cover database schema, FHIR mappings, and pre-built SQL query templates.360installs4Firecrawl Researchfirecrawl-research is an agent skill from glebis/claude-skills that uses Firecrawl to crawl and extract structured content from web pages during early product research. Developers invoke it when they need competitor feature comparisons, market landscape summaries, or sourced answers to validation questions without manually copying pages. The skill orchestrates Firecrawl crawl and scrape calls so an agent can synthesise findings from multiple URLs into research-ready notes. It fits the window before a validation prototype or scoped build, when teams still need external evidence about pricing, positioning, and feature gaps. Output is extracted page content and agent-synthesised research summaries grounded in crawled sources.354installs5Presentation Generatorpresentation-generator is an Office & Documents skill from glebis/claude-skills that drafts slide decks for pitch, roadmap, customer, and investor narratives. The skill produces structured outlines, speaker notes, and export-ready presentation content so developers and technical leads can communicate product direction without starting from a blank deck template. Developers reach for presentation-generator when they need a coherent narrative arc—problem, solution, milestones, and ask—formatted for slides before a customer review, roadmap sync, or investor conversation during active product build cycles.292installs6Deep Researchdeep-research is a Claude agent skill from glebis/claude-skills built for structured technical and market investigation before implementation commitments. It runs layered queries across APIs, regulations, competitive landscapes, and architectural alternatives, then synthesizes findings with citations developers can trace. The skill fits early discovery when choosing vendors, compliance paths, integration strategies, or build-versus-buy decisions. Developers reach for deep-research when a single web search is insufficient and they need an agent to compare options, surface risks, and produce a decision-ready brief instead of ad-hoc notes scattered across tabs.288installs7Decision Toolkitdecision-toolkit is a glebis/claude-skills agent skill that turns ambiguous choices into structured decision support artifacts rather than making the call for you. It ships seven frameworks—First Principles, 10-10-10, Pre-Mortem, Regret Minimization, Weighted Decision Matrix, Opportunity Cost Calculator, and Scenario Matrix—plus a bias encyclopedia covering 20+ cognitive biases with counter-questions. Outputs include interactive HTML wizards, Markdown decision records, and voice-summary templates. Invoke with /decision-toolkit plus a decision question when facing technology pivots, architecture forks, or significant engineering investments that need documented criteria and tradeoff visibility.287installs8Elevenlabs Ttselevenlabs-tts is a glebis/claude-skills agent skill that integrates ElevenLabs text-to-speech into software development workflows when teams need spoken audio from written content. The skill guides voice selection, script preparation, and API usage so developers turn UI strings, agent replies, demo narration, or podcast drafts into natural-sounding speech without building a custom audio pipeline from scratch. Developers reach for elevenlabs-tts when prototyping voice-enabled features, adding accessibility read-aloud behavior, generating demo audio for presentations, or piping LLM responses to audio output in agent products. It fits mobile apps, web dashboards, and AI agents that call the ElevenLabs TTS API during build and iteration. The skill emphasizes practical integration steps and content formatting so generated audio is production-ready for tests, stakeholder demos, or early voice UX validation.276installs9Telegramtelegram is an AI & Agent Building agent skill from glebis/claude-skills that connects Claude Code to Telegram bots for development-time communication. It supports build alerts pushed to a chat, slash-command handlers that trigger agent actions, and async human-in-the-loop messaging when an agent needs approval or context mid-run. Developers reach for telegram when long-running coding agents should notify a channel on completion, accept commands from mobile, or pause for human review without staying in the terminal. The skill bridges Telegram Bot API plumbing and Claude Code session control so agent workflows extend beyond the local shell into a messaging surface teams already monitor.249installs10Brand AgencyThe brand-agency skill from glebis/claude-skills acts like a lightweight brand consultancy for early product work. It helps teams articulate positioning, audience fit, tone of voice, and visual identity direction so SaaS, ecommerce, and content products present a coherent brand before UI, marketing, or go-to-market execution begins.241installs11Doctorgdoctorg organizes and produces technical documentation for software projects, turning scattered notes into coherent READMEs, architecture notes, and operational guides. It supports agent-assisted and human teams building SaaS or API products who need documentation that stays accurate as code evolves.230installs12Transcript AnalyzerThe transcript-analyzer skill from glebis/claude-skills uses LLM-style analysis to turn raw meeting, interview, and call transcripts into structured research output. It highlights themes, decisions, objections, and follow-ups so agent, SaaS, and content teams can accelerate discovery, customer insight synthesis, and downstream planning without manually re-reading long recordings.228installs13GmailCovers Gmail integration patterns in glebis/claude-skills for authenticating users, calling Gmail APIs, and automating inbox actions so SaaS products and agents can send alerts, parse messages, and run email-driven workflows safely.227installs14ZoomImplements focal zoom and pan patterns for data-dense visualizations, enabling readers to explore local detail on demand while preserving a clean macro view aligned with information-design best practices.223installs15Fathomfathom connects Fathom meeting capture to Claude Code so recorded calls become searchable summaries, decisions, and tasks. It suits SaaS and agent operators who run frequent customer or internal meetings and need reliable post-call processing without manual note rewriting.219installs16TddThe tdd skill instructs Claude Code to follow test-driven development by writing failing tests first, implementing minimal passing code, and refactoring safely, giving teams a disciplined build-time workflow that locks behavior before features expand across backend services and CLIs.219installs17Github GistThe github-gist skill teaches Claude Code to create, update, and manage GitHub Gists so small files and text bundles can be published as shareable URLs during development, ideal for repro steps, config examples, and lightweight documentation outside the main repository.217installs18Youtube TranscriptDelivers the youtube-transcript skill from glebis/claude-skills, fetching and normalizing YouTube captions so agents can search, summarize, and extract insights from video content during early discovery, research, and content-planning workflows.217installs19Llm CliThe llm-cli skill documents how Claude Code should drive command-line LLM tools for ad hoc inference, batch jobs, and prompt experiments, giving teams a repeatable shell pattern for model access during agent-tooling and backend prototyping work.206installs20BalancedThe balanced skill from glebis/claude-skills configures a measured assistant stance for Claude workflows. It promotes nuanced, evidence-aware guidance—surfacing alternatives, risks, and uncertainties—so agents remain helpful during implementation without overselling fragile solutions or skipping validation steps.204installs21GwsThe gws skill guides Claude Code through Google Workspace integration patterns so agents can query or act on Gmail, Calendar, Drive, and related services while building features that depend on real user productivity data instead of static mocks.204installs22Chrome Historychrome-history from glebis/claude-skills is an idea-phase research skill specialized for Google Chrome. It guides agents to locate, query, and summarize Chrome history databases so prior browsing becomes structured evidence for specs, competitor review, or personal knowledge capture.203installs23Metameta helps teams ship correct HTML metadata—titles, descriptions, social cards, and canonical links—so new pages index cleanly and render compelling previews when distributed at launch.199installs24Granolagranola bridges Granola AI meeting notes into Claude workflows, transforming conversational highlights into structured updates for specs, tickets, and team memory. It targets SaaS builders and agent operators who depend on fast, accurate meeting capture during day-to-day iteration.198installs25Context Buildercontext-builder prepares agent-ready context packs by selecting relevant code, docs, tickets, and constraints before Claude Code work begins. It targets teams shipping SaaS or API products where agents must ground answers in real repository state instead of guessing paths or APIs.196installs26Wispr AnalyticsWispr-analytics from glebis/claude-skills guides Claude Code through querying, interpreting, and acting on Wispr speech-dictation analytics for adoption, engagement, and retention after features ship.187installs27Nano Banananano-banana equips Claude Code to call Nano Banana-style image models from agent sessions, standardizing prompts, sizes, and retries so teams can produce marketing art, UI placeholders, and iteration sets without a separate design toolchain.179installs28Gpt Image 2gpt-image-2 wires GPT-class image generation into Claude Code so teams can create and refine visuals—mockups, icons, hero art, and social assets—during development without external design tools.173installs29Temple GeneratorTemple-generator from glebis/claude-skills procedurally produces temple structures, motifs, and scene-ready descriptions for games, creative prototypes, and narrative content workflows inside Claude Code.170installs30Meeting Processormeeting-processor turns raw meeting transcripts into structured outputs—summaries, decisions, risks, and actionable tasks—for product and engineering teams. It supports SaaS and agent operators who hold frequent syncs and need consistent post-meeting artifacts without manual reformatting across tools.164installs31Session SearchThe session-search skill enables Claude Code to query full-text content inside archived agent sessions so teams can retrieve earlier explanations, stack traces, and implementation notes during ongoing operation and iterative maintenance work.163installs32Browsing Historybrowsing-history from glebis/claude-skills is an idea-phase research skill that reads browser visit data to rebuild context. Agents use it to summarize what was already explored, cite prior sources, and accelerate early discovery without repeating manual web searches.159installs33Jtbdjtbd applies the Jobs-to-be-Done framework in Claude Code to turn raw ideas into validated job statements, interview scripts, and scoped outcomes that anchor prototype and landing decisions.155installs34Recordingrecording standardizes how Claude Code documents software behavior—capturing sessions, annotating repro steps, and packaging clips for QA sign-off, release communication, and user onboarding.155installs35Tufte ReportBuilds publication-quality Tufte-inspired reports that prioritize information density, intentional typography, small multiples, and low chartjunk for communicating analyses and product metrics to technical and business readers.155installs36Lab Retrolab-retro structures team or lab retrospectives into clear themes, learnings, and owned follow-ups after a delivery cycle. It helps SaaS and agent teams running regular sprints or experiments institutionalize improvement without ad hoc retro docs that lose momentum between sessions.153installs37Vision Benchvision-bench equips Claude Code to run structured vision model evaluations—accuracy, latency, cost, and failure modes—on fixed image sets so teams can ship multimodal features with measured quality instead of anecdotal prompts.146installs38Skill Studioskill-studio is an in-repo workshop for building Claude Code skills—structuring SKILL.md files, bundling references and scripts, and iterating agent capabilities with consistent progressive-disclosure patterns.145installs39Linearlinear connects Claude Code to Linear for hands-on project management—creating issues, updating states, and keeping engineering work aligned with cycles, labels, and team workflow conventions.141installs40Session FinderThe session-finder skill helps Claude Code discover specific historical coding sessions using metadata and filesystem cues, enabling operators to reopen the right conversation, continue interrupted work, and audit past agent actions without manually browsing log directories.136installs41Timebuzzer LedThe timebuzzer-led skill from glebis/claude-skills supports building integrations with Timebuzzer LED time-tracking hardware. It helps connect button presses and LED feedback to software timers, project labels, and workflow automations so SaaS, API, and agent products capture accurate activity data with less manual overhead.136installs42Session AnonymizerThe session-anonymizer skill teaches Claude Code to sanitize coding session transcripts by removing or masking PII, API keys, and other sensitive strings so teams can share debugging history externally while meeting security and compliance expectations.126installs43Sortedsorted helps Claude Code apply disciplined ordering to tasks, issues, and notes—using priority, status, and custom keys—so product and engineering backlogs stay legible during weekly iteration without manual spreadsheet shuffling.122installs44CodexThe codex skill from glebis/claude-skills emulates disciplined Codex-like agent coding inside Claude. It choreographs planning, targeted edits, and verification passes so complex changes land safely across services, scripts, and UI code while preserving context, testability, and review-friendly diffs.119installs45Rag EvalA developer tool for AI integration and automation. This is a developer tool for building and integrating AI-powered features.118installs46Tg ResponderReads a responder queue database to present pending reply drafts for approve/edit/skip, sends approved ones via the telegram skill, and tracks follow-ups. A developer uses it to triage a Telegram inbox and chase messages that never got a reply.116installs47Wow DigestIngests the last 24h of newsletters and Telegram posts, filters noise, and scores survivors for genuine surprise against the user's focus before appending WOW items to today's note. A developer uses it for morning reading that surfaces surprising content, not just relevant content.116installs48Cognitive ToolkitProvides evidence-based CBT and DBT intervention workflows (7-step thought record, Opposite Action, distortion detection) delivered inside Claude Code with configurable pushback. A developer uses it for terminal-native, local, privacy-first mental-health self-help exports importable into a PKM.115installs49RetrospectiveScans the current or a day's worth of session transcripts to extract user corrections, skill failures, and patterns, then proposes concrete actions like skill edits and memories. A developer uses it to wrap up a session and capture learnings into their setup.104installs50Site DiagnosisWalks a client through five rounds of AskUserQuestion prompts about their project, tools, pain points, and goals, then produces a structured consultation brief. A developer or consultant uses it to prepare for a website review with a client whose AI-built site has problems.87installs51EcosystemRuns four sequential checks over skills, projects, CLAUDE.md files, and a remote Mac Mini, then prints a report with an attention rollup and appends a summary to the daily note. A developer uses it on demand to see what is stale, broken, or down across their setup.86installs52PresentTurns research or notes into one index.html with a readable article mode and a navigable narrated slides mode, using ElevenLabs TTS and configurable detail levels. A developer uses it to produce a shareable narrated briefing or deck from content.85installs53Learning VaultRuns an interactive setup to capture the goal, domains, and self-assessment, then builds a full Obsidian vault with templated notes, wikilinks, tasks, and a working bundled Dataview plugin. A developer uses it to spin up a certification-prep or study knowledge base with priority ordering and progress queries.73installs54Weekly DigestRuns a seven-phase pipeline (scoping, research, verification, scoring, output, presentation) that scores candidates on five parameters including slopiness and produces markdown plus a Tufte-style HTML report. A developer uses it to monitor a topic with quality filtering and publish a curated roundup.66installs55Agency Meetup PublishAn end-to-end pipeline that publishes AGENCY Community meetup recordings to YouTube by downloading the Zoom recording, adding intro/outro, generating a thumbnail and timecoded description, uploading, and adding to a playlist. A developer uses it to process recorded meetups into published videos.64installs56Qmd SearchWraps the qmd engine to semantically search a markdown knowledge base on-device, matching meaning across languages and offering keyword, vector, hybrid, and literal grep modes. A developer uses it to find notes by concept or cross-lingual query without anything leaving the machine.63installs57Publish SkillAutomates the full pipeline of syncing a skill's SKILL.md into the site repo, generating MDX metadata and a hero image, and committing/pushing both repos. A developer uses it to add a new skill to the site or update an existing one while preserving activity data.59installs58Agency SocialsGenerates event covers, YouTube thumbnails, and social post images for the AGENCY Community using an established design system of colors, fonts, and layout rules. A developer uses it to produce on-brand visual assets at defined sizes like 800x800 and 1280x720.58installs59Meeting PrepFetches upcoming Cal.com bookings, researches each participant, mines vault history, and writes structured prep notes (solo or group) into Obsidian. A developer uses it to prep for upcoming calls and later link prep notes to their post-meeting session notes.58installs60Coaching Session SummarizerReads a synced Fathom/Granola transcript, analyzes it directly on the subscription with no API call, and appends key insights, decisions, action items, and optional deep cross-session analysis. A developer uses it after a session syncs to their Obsidian vault to get a structured summary linked to trails.56installs61Synthetic Session GeneratorProduces fictional but believable practitioner/client dialogue grounded in a chosen modality with persona consistency and modality fidelity, watermarked as synthetic. A developer uses it to build eval datasets, product demos, or few-shot examples for session-analysis tools without real client data.56installs62CullMaps the Cull image-library app's automation surface (CLI, cull:// URL scheme, optional MCP) for importing folders, rating, building collections, quality analysis, find-similar, and export. A developer uses it to review and organize image batches headlessly, always displaying them in Cull rather than opening files.55installs63Agent CliConverts Python CLI scripts to emit agent-consumable NDJSON via a --json flag, or scaffolds a complete cli_utils package with helpers, tests, and license. A developer uses it to make scripts machine-readable for AI agents or to build an open-source CLI-for-agents library.52installs64Name AuditionGenerates candidate names then screens them via a domain-availability script and web collision research across SaaS, GitHub, packages, and the adjacent vertical, producing a ranked risk table. A developer uses it to pressure-test a naming shortlist before committing, catching adjacent-domain collisions a domain check alone misses.52installs65Agency Docs UpdaterAn end-to-end pipeline that publishes Claude Code lab meetings by finding/creating a Fathom transcript, downloading video, uploading to YouTube, generating a fact-checked Russian summary, and creating MDX docs. A developer uses it to replace 5+ manual steps in publishing recorded meetings.51installs66Job BabysitterStarts one background watcher that polls with backoff, detects plateaus, and distinguishes done from stuck, emitting a verdict of done, needs-attention, or blocked. A developer uses it to stop manually polling long jobs and get routed the exact next step when one completes or hangs.50installs67Library SyncHandles adding, translating, and updating MDX library articles in en/ and ru/ folders, checking sync status and running a Russian style review. A developer uses it to keep an agency-docs bilingual library consistent between English and Russian versions.50installs68Peer Agent CollaborationDefines a parity-based operating model for multi-agent work: choosing the lead by operating surface, a handoff template, a challenge protocol, and high-value workflows. A developer uses it to structure how two coding or business agents split, hand off, and challenge each other's work.50installs69VaultRuns read-only probes to check each of three independent locks and prints the exact command to fix any gap, never moving or encrypting data itself. A developer uses it to verify a defense-in-depth storage posture for real therapy or coaching transcripts.44installs70Browser MateLaunches or reuses a per-profile debug Chrome instance with its own user-data-dir and port so agent-browser can drive authenticated sessions non-destructively. A developer uses it to automate sites like ChatGPT or LinkedIn while preserving their existing browser windows.43installs71AnonRuns a local layered detector (regex, Natasha, local LLM) over transcripts and replaces PII with unique reserved-sentinel placeholders, emitting a redacted GREEN copy and a secret local map.json. A developer uses it before sharing a transcript or sending it to a cloud LLM, keeping raw text and originals on the machine.42installs72RedRe-runs the CONFIDE detectors on redacted output to flag singling-out, linkability, and optional inference risk mapped to GDPR Art-29, reporting categories and counts only. A developer uses it after redacting to gauge what an attacker could still infer before sharing.41installs73AuditRuns the local layered detector stack over every file in a folder and emits only aggregates (counts by type and layer, per-session redaction-rate distribution, a residual proxy). A developer uses it to size PII across a raw corpus or check residual leakage across an already-redacted corpus.40installs74RehydrateReplaces CONFIDE placeholders in a cloud-produced analysis with their originals from a local secret map, tolerating mangled placeholder forms and flagging unmatched ones. A developer uses it as the final step after analyzing redacted text in the cloud without exposing originals.40installs75ReleaseA config-driven release orchestrator that enforces SemVer against declared stable surfaces, runs a readiness gate, drafts a Keep-a-Changelog section, and tags the release. A developer uses it to ship a version with compatibility tiers and deprecations tracked.40installs76SetupOne-shot installer and config writer that sets up CONFIDE's dependencies (natasha, scrubadub, phonenumbers, pymorphy2), pulls the Ollama model, and writes defaults so anon and red work with no further config. A developer runs it first to get local PII de-identification ready.40installs77Rigorous ExperimentsEnforces rigorous n-of-1 statistical practice (pre-registration, exact permutation, fixed-family BH, stationarity and Simpson checks, adversarial review) across design, conduct, validate, and audit modes. A developer uses it to test whether a correlation in health, behavioral, or self-tracking data is real.36installs78Init Tauri AppDelegates boilerplate to npm create tauri-app then layers an opinionated core and selected modules, gating on cargo check and npm build after each step. A developer uses it to start a new Tauri desktop project with proven conventions and optional JTBD product context.24installs79Whitepaper AuditRuns two lanes over a document (script checks for readability, acronyms, structure, links; an LLM judge for overclaims, inconsistent numbers, and audience fit) and merges them into a prioritized findings report. A developer uses it to QA a technical white paper before publishing, optionally applying fixes.24installs80Init Xcode AppGenerates a native SwiftUI project from project.yml (committed .xcodeproj) with swiftformat/swiftlint, optional CloudKit/CI/Release/Push modules, and a build+test gate. A developer uses it to start a new Mac/iOS/watchOS app with the loop-closer toolchain preloaded.23installs81Repo PrepWalks a repo through scaffolding the legal, metadata, community, and compliance files needed before open-sourcing, driven from a central author profile and confirmed per repo. A developer uses it to prep a new or existing repo for GitHub, including an AUTHORSHIP record for AI-assisted work.22installs