921 mods tagged github that install into Claude Code, measured the same four ways as everything else here: what each costs in tokens on every session, what a security scan found and why, whether it is the original, and what changed.
Schreibe (oder überarbeite) Text so, dass er nach Holger klingt — dem IT-Freelancer hinter codingsoul.org — und NICHT nach der polierten agent-smith.org-Landingpage. Nutze diesen Skill immer, wenn Text für Blogposts, READMEs, Doku, Release-Notes, Issues, PR-Beschreibungen oder Marketing rund um Agent Smith /…
Comprehensive GitHub Pull Request code review skill. Use when asked to "review this PR", "code review", "review pull request", "check this PR", or when a GitHub PR URL is provided. Fetches PR metadata, diff, comments, commits, and related issues using gh CLI. Creates organized review workspace, analyzes code against…
Web search, content extraction, and research tools for Claude Code. Search the web, Reddit, Hacker News, GitHub, X/Twitter, Wikipedia, arXiv, and more — all from your Claude session.
★not rated 24 21d agoA
tokens not measured
originalMIT
Analyze any Git repository in depth — understand architecture, debug across repo boundaries, review APIs, plan migrations, and evaluate libraries. Surfaces real function signatures, parameter types, return values, and source citations with exact file paths and line numbers. Eliminates hallucinated APIs by grounding…
Map Azure architectures and operations to security and compliance controls. Use this skill when users ask for governance, policy enforcement, audit readiness, or regulatory alignment. Covers ISO, SOC2, HIPAA, PCI-DSS, and custom compliance frameworks.
Adversarial code review for Claude Code: each GitHub PR or GitLab MR runs a gauntlet of up to seven specialist agents for bugs, security, tests and cross-file impact, and every finding must survive verification, a skeptical validator, and a blind challenge before it is posted.
★not rated 13
changed 3d agoA
tokens not measured
originalApache-2.0
Enrich an existing PKUP report directory — group commits into tasks, filter out non-creative groups, and write Polish descriptions. Use when the user has already generated a PKUP report and wants to clean it up or add "Zaprojektowałem..." descriptions.
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
Generate or update the architecture documentation in docs/content/architecture/. Use on "update architecture docs", "generate architecture documentation", "regenerate architecture docs", or after any structural change to the codebase.
Research and stress-test open-source project names, from collision checks to launch-ready decisions. Use when the user asks to name or rename a project, compare candidate names, screen GitHub or package namespace risk, verify language or mythology claims, or improve SEO/GEO clarity. Perform live web and registry…
The most visual MCP server for the Google Ads Transparency Center. Ask about any brand and your AI shows actual ad copy rendered in chat. Runs locally from the @proxy-intell/google-ads-library-mcp npm package.
★not rated 10 2d agoA
tokens not measured
originalMIT
Turn a GitHub pull request into an interactive review map with AI insights, reviewed in the browser. Use when the user asks to review, map, or visualize a PR.
Classify your GitHub starred repositories into semantic categories and sync them to GitHub Lists, using Claude's reasoning instead of static keyword rules.
★not rated 9 6mo agoA
tokens not measured
originalMIT
Regelmäßiger Pflegedurchlauf für ein bereits veröffentlichtes GitHub-Repository (Stufe 1, günstig und oft wiederholbar): zuerst alle Distributionsflächen des Projekts ermitteln (npm, PyPI, Registries, Marketplaces, Stores, Website) und Änderungen später dorthin spiegeln, dann Topics setzen, Privacy-Gate, Dokumente auf…
Use this skill whenever the user wants to operate on git worktrees — create, enter, list, remove, or garbage-collect — including parallel agent sessions, side-by-side PR reviews, benchmarking multiple branches, or isolating a dirty tree from a fresh checkout. Trigger even when "worktree" is not said — phrasings like…
★not rated 7 8d agoA158 tokens
originalMIT
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: