AI Plugin Radar

Claude & Codex Ecosystem

100 curated repos · 100 reviewed by Claude · 100 new this scan · updated 2026-06-03

How to read this. The verdict marked ✦ is Claude reading each repo's README and key files (skills, MCP configs, hooks, install scripts) and flagging concerns. Repos not yet reviewed fall back to a heuristic indicator. Neither is a formal security audit and either can be wrong — always review code yourself before installing or running anything.

100Repositories

100Claude-reviewed

56Lower risk

39Moderate

5Elevated

0Higher risk

New only

Repository	Stars	Forks	Language	License	Updated	Risk
affaan-m/ECCNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	204168	31317	JavaScript	MIT	2026-06-02	✦ Moderate
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond. A large 'agent harness' bundle of skills, hooks, and commands for Claude Code and other agents (memory, instincts, research-first workflows). ai-agentsanthropicclaudeclaude-codedeveloper-toolsllmmcpproductivity ✦ Claude review — Moderate The sampled skills (introspection, API design, article writing) are legitimate instructional content with no injection or exfiltration. Risk comes from sheer surface area, not malice: 94 shell/install scripts and an npm installer mean a lot of code runs at install time that no one reviews line by line. The 200k+ star count is implausibly high for a repo like this, so treat popularity as a weak trust signal here. MED 94 install/shell scripts plus an npm installer (ecc-universal) = large unreviewed install-time surface LOW Star/fork counts look inflated; don't lean on popularity as a safety signal Heuristic signals Popular: 204,168 stars Licensed (MIT) Active (updated 0d ago) 31317 forks Dependency manifests: 5 Contains install/shell scripts (94) Has CI workflows Has SECURITY.md
multica-ai/andrej-karpathy-skillsNEW PluginsSkills	166016	16990	—	None	2026-04-20	✦ Lower risk
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls. A single CLAUDE.md / one skill of behavioral guidelines to curb common LLM coding mistakes (think before coding, surgical changes). ✦ Claude review — Lower risk Pure instructional text, no scripts, no tool permissions, nothing executable. The only gap is a missing repo-level license (the skill frontmatter says MIT but the repo has none). LOW No clear repo license despite MIT in skill frontmatter Heuristic signals Popular: 166,016 stars No clear license Updated 43d ago 16990 forks
x1xhlol/system-prompts-and-models-of-ai-toolsNEW General / Tooling	138706	34510	—	GPL-3.0	2026-05-23	✦ Lower risk
FULL Augment Code, Claude Code, Cluely, CodeBuddy, Comet, Cursor, Devin AI, Junie, Kiro, Leap.new, Lovable, Manus, NotionAI, Orchids.app, Perplexity, Poke, Qoder, Replit, Same.dev, Trae, Traycer AI, VSCode Agent, Warp.dev, Windsurf, Xcode, Z.ai Code, Dia & v0. (And other Open Sourced) System Prompts, Internal Tools & AI Models A collection of leaked/extracted system prompts and internal tool configs from many commercial AI products. Reference material, not code. aiboltcluelycopilotcursorcursoraidevingithub-copilotlovableopen-source ✦ Claude review — Lower risk No executable surface, so little technical risk to an agent reading it. Caveats are non-security: the content is other companies' material of uncertain provenance, and the README solicits crypto donations. Use as curiosity/reference, not as a trusted source. INFO Reference text only; no scripts or skills to execute LOW README solicits crypto donations; content provenance is uncertain Heuristic signals Popular: 138,706 stars Licensed (GPL-3.0) Active (updated 10d ago) 34510 forks
anthropics/claude-codeNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	129651	21080	Python	None	2026-06-02	✦ Lower risk
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands. The official Claude Code CLI from Anthropic, plus its bundled first-party plugins and skills. ✦ Claude review — Lower risk First-party, actively maintained, with a SECURITY.md. The heuristic dinged it for a curl\|bash installer and 'no license', but the curl\|bash line is the official documented installer and the repo ships under Anthropic's own terms. Lowest practical risk of this set. INFO curl\|bash install is the official, documented installer Heuristic signals Popular: 129,651 stars No clear license Active (updated 0d ago) 21080 forks README uses curl\|bash style install Contains install/shell scripts (21) Has CI workflows Has SECURITY.md
garrytan/gstackNEW HooksMCP ServersSkillsSubagents	106361	15828	TypeScript	MIT	2026-06-01	✦ Moderate
Use Garry Tan's exact Claude Code setup: 23 opinionated tools that serve as CEO, Designer, Eng Manager, Release Manager, Doc Engineer, and QA Garry Tan's opinionated 23-tool Claude Code setup (CEO/designer/eng-manager review skills, a headless browser QA skill, benchmarking). ✦ Claude review — Moderate Skills are legitimate and well-structured. The notable behavior: most skills carry a 'Preamble (run first)' that auto-executes a bash update-check (gstack-update-check) on activation. That's benign as written but means activating a skill silently runs a script and likely phones home for updates. 26 shell scripts overall. MED Skills auto-run a bash 'preamble' (gstack-update-check) on activation, likely a network update check LOW 26 install/shell scripts Heuristic signals Popular: 106,361 stars Licensed (MIT) Active (updated 1d ago) 15828 forks Dependency manifests: 1 Contains install/shell scripts (26) Has CI workflows Has SECURITY.md
farion1231/cc-switchNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	89515	5837	Rust	MIT	2026-06-02	✦ Moderate
A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Gemini CLI & Hermes Agent. Only official website: ccswitch.io A cross-platform Tauri desktop app for managing provider configs/API keys across Claude Code, Codex, Gemini CLI and others. ai-toolsclaude-codecodexdesktop-apphermeshermes-agentmcpminimaxomoopen-source ✦ Claude review — Moderate Legitimate config manager (Rust/Tauri + React). By design it reads, stores, and switches provider API keys, so it handles secrets on disk — fair to use, but understand it holds your keys. No malicious patterns in sampled code. Distributed as desktop binaries, so trust the release channel. MED By design stores and manages provider API keys on disk LOW Ships prebuilt desktop binaries; verify the release source Heuristic signals Popular: 89,515 stars Licensed (MIT) Active (updated 0d ago) 5837 forks Dependency manifests: 2 Contains install/shell scripts (5) Has CI workflows Has SECURITY.md
nextlevelbuilder/ui-ux-pro-max-skillNEW PluginsSkillsSlash Commands	86567	8938	Python	MIT	2026-04-03	✦ Moderate
An AI SKILL that provide design intelligence for building professional UI/UX multiple platforms A large design-intelligence skill pack (banners, brand, design systems, slides) with a CLI installer. ai-skillsantigravityclaudeclaude-codecodexcommand-linecopilotcursor-aihtml5kiro ✦ Claude review — Moderate Content is design guidance plus helper Node scripts (e.g. inject-brand-context.cjs). No injection or exfiltration seen. Some skills shell out to bundled scripts and it asks for PayPal support; it's been 60 days since update. Standard 'review the bundled scripts before running' caution. LOW Skills invoke bundled Node scripts; review before running INFO Last updated ~60 days ago Heuristic signals Popular: 86,567 stars Licensed (MIT) Updated 60d ago 8938 forks Dependency manifests: 3 Contains install/shell scripts (1) Has CI workflows
thedotmack/claude-memNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	80291	6913	TypeScript	Apache-2.0	2026-05-29	✦ Moderate
Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More Persistent cross-session memory for agents: captures session activity, compresses it with AI, and re-injects context later. aiai-agentsai-memoryanthropicartificial-intelligencechromadbclaudeclaude-agent-sdkclaude-agentsclaude-code ✦ Claude review — Moderate Plausible, useful design and the orchestration skills are clean. Two things warrant care: the documented install is a piped 'curl ... \| bash' one-liner that can take your API key as an argument, and the tool by nature records and stores everything your agent does. 43 shell scripts. MED Install is curl\|bash, and an option pipes your API key on the command line LOW By design captures/stores full session activity Heuristic signals Popular: 80,291 stars Licensed (Apache-2.0) Active (updated 4d ago) 6913 forks README uses curl\|bash style install Dependency manifests: 3 Contains install/shell scripts (43) Has CI workflows Has SECURITY.md
JuliusBrussee/cavemanNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	67976	3827	JavaScript	MIT	2026-05-20	✦ Moderate
🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman A skill/plugin that compresses agent output into terse 'caveman' phrasing to cut tokens, with subagent presets. aianthropiccavemanclaudeclaude-codellmmemeprompt-engineeringskilltokens ✦ Claude review — Moderate Harmless premise and the cavecrew subagent skills are just delegation guidance (duplicated across many agent ecosystems' folders). Risk is the usual curl\|bash installer and 16 shell scripts; nothing malicious in the sampled content. MED curl\|bash style install LOW 16 install/shell scripts Heuristic signals Popular: 67,976 stars Licensed (MIT) Active (updated 13d ago) 3827 forks README uses curl\|bash style install Dependency manifests: 5 Contains install/shell scripts (16) Has CI workflows Has SECURITY.md
shareAI-lab/learn-claude-codeNEW HooksMCP ServersPluginsSkillsSubagents	64320	10511	Python	MIT	2026-06-02	✦ Lower risk
Bash is all you need - A nano claude code–like 「agent harness」, built from 0 to 1 An educational 'nano Claude Code' agent harness built from scratch, with teaching skills (agent-builder, code-review, mcp-builder, pdf). agentagent-developmentai-agentclaudeclaude-codeeducationalllmpythonteachingtutorial ✦ Claude review — Lower risk Primarily a learning resource; the skills are instructional and the bash snippets inside them are conventional (npm audit, pdftotext). MIT licensed, active, no install one-liner flagged. Low risk as reference/learning material. INFO Educational content; sampled skills contain only conventional shell examples Heuristic signals Popular: 64,320 stars Licensed (MIT) Active (updated 0d ago) 10511 forks Dependency manifests: 2 Has CI workflows
gsd-build/get-shit-doneNEW HooksMCP ServersSkillsSlash CommandsSubagents	63847	5435	JavaScript	MIT	2026-05-31	✦ Moderate
A light-weight and powerful meta-prompting, context engineering and spec-driven development system for Claude Code by TÂCHES. A meta-prompting / spec-driven development system for Claude Code (GSD). This repo is archived and redirects to open-gsd/gsd-core. claude-codecontext-engineeringmeta-promptingspec-driven-development ✦ Claude review — Moderate The README is now just a 'we moved' redirect, so installing from here gets a stale snapshot — go to the active repo instead. The changesets describe a PostToolUse hook that fires after git ops and dispatches work in a detached subprocess; benign but it's an auto-running, backgrounded hook. 66 shell scripts. MED Repo archived/moved to open-gsd/gsd-core; this is a stale redirect MED Bundles a PostToolUse hook that auto-runs after git ops and spawns a detached subprocess Heuristic signals Popular: 63,847 stars Licensed (MIT) Active (updated 2d ago) 5435 forks Dependency manifests: 2 Contains install/shell scripts (66) Has CI workflows Has SECURITY.md
ComposioHQ/awesome-claude-skillsNEW HooksMCP ServersPluginsResource ListsSkillsSlash Commands	62986	6917	Python	None	2026-05-22	✦ Lower risk
A curated list of awesome Claude Skills, resources, and tools for customizing Claude AI workflows A large curated list of Claude skills and resources, with a few sample skills bundled (artifacts-builder, brand-guidelines, canvas-design). agent-skillsai-agentsantigravityautomationclaudeclaude-codecodexcomposiocursordeveloper-tools ✦ Claude review — Lower risk Mostly a curated index plus example skills that mirror Anthropic's own. No injection seen. As with any 'awesome' list, the real risk lives in the third-party links you follow — vet each linked skill on its own. No repo license is set. LOW Aggregates many third-party skills; vet linked items individually LOW No clear repo license Heuristic signals Popular: 62,986 stars No clear license Active (updated 11d ago) 6917 forks Dependency manifests: 2 Contains install/shell scripts (2) Has CI workflows
safishamsi/graphifyNEW HooksMCP ServersSkills	58467	6107	Python	MIT	2026-06-02	✦ Moderate
AI coding assistant skill (Claude Code, Codex, OpenCode, Cursor, Gemini CLI, and more). Turn any folder of code, SQL schemas, R scripts, shell scripts, docs, papers, images, or videos into a queryable knowledge graph. App code + database schema + infrastructure in one graph. Turns a folder of code/docs/media into a queryable knowledge graph for coding agents; outputs interactive HTML and JSON. antigravityclaude-codecodexgeminigraphragknowledge-graphleidenopenclawragskills ✦ Claude review — Moderate Legitimate tooling. It installs an optional git post-commit/post-checkout hook (Python that appends to existing hooks) to auto-rebuild the graph, and the /graphify command can clone a remote repo you point it at. The sample .mcp.json uses a clearly labeled placeholder token (good practice). Curl\|bash install. Nothing malicious. MED Installs git post-commit/post-checkout hooks that auto-run; curl\|bash install INFO Sample MCP config uses an obvious placeholder token, not a real secret Heuristic signals Popular: 58,467 stars Licensed (MIT) Active (updated 0d ago) 6107 forks README uses curl\|bash style install Dependency manifests: 3 Contains install/shell scripts (8) Has CI workflows Has SECURITY.md
nexu-io/open-designNEW HooksMCP ServersPluginsResource ListsSkillsSlash CommandsSubagents	57694	6513	TypeScript	Apache-2.0	2026-06-03	✦ Moderate
🎨 Local-first, open-source Claude Design alternative. 🖥️ Native desktop app. ⚡ 259+ Skills · ✨ 142+ Design Systems 🖼️ Web · desktop · mobile prototypes · slides · images · videos · HyperFrames 📦 Sandboxed preview · HTML/PDF/PPTX/MP4 export 🤖 Claude Code / OpenClaw / Codex / Cursor / OpenCode / Qwen / Copilot / Hermes / Kimi & 17+ CLIs. A local-first open-source 'Claude Design' alternative: native desktop app with 250+ skills and design systems, sandboxed preview, multi-format export. agent-skillsai-agentsai-designbyokclaude-code-for-designclaude-designcodex-designcoding-agentscursor-designdesign-systems ✦ Claude review — Moderate Big, active, Apache-licensed project. Skills are well-scoped (some declare allowed-tools like Bash/WebFetch, which is appropriate transparency). Risk is scale: 61 shell scripts, 22 dependency manifests, a desktop app, and a built-in model router. No injection seen; treat it like any large desktop tool. MED Large surface: desktop app, 61 shell scripts, 22 manifests, built-in model router INFO Skills declare allowed-tools explicitly (good transparency) Heuristic signals Popular: 57,694 stars Licensed (Apache-2.0) Active (updated 0d ago) 6513 forks README uses curl\|bash style install Dependency manifests: 22 Contains install/shell scripts (61) Has CI workflows
ruvnet/rufloNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	57549	6578	TypeScript	MIT	2026-06-02	✦ Elevated
🌊 The leading agent meta-harness for Claude. Deploy intelligent multi-agent swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning swarm intelligence, RAG integration, and native Claude Code / Codex Integration A multi-agent 'swarm' meta-harness for Claude with adaptive coordination, memory, and — notably — an autonomous payments module. agentic-aiagentic-frameworkagentic-ragagentic-workflowagentsai-agentsai-assistantai-codingai-skillsautonomous-agents ✦ Claude review — Elevated Highest-impact surface in this batch. Coordinator skills embed 'hooks.pre' blocks that auto-execute shell and mcp__claude-flow__ calls on activation, and the bundle includes an 'agentic-payments' agent that signs transactions (Ed25519) and authorizes autonomous AI purchases. Combined with 179 install scripts and 56 dependency manifests, this is a lot of auto-running, high-consequence machinery. Nothing here is proof of malice, but autonomous payment authorization plus auto-running hooks is exactly where you'd want a careful human review and tight spend caps before installing. HIGH Bundles an autonomous-payments agent that cryptographically signs and authorizes AI purchases MED Coordinator skills auto-run shell + MCP calls via hooks.pre on activation MED 179 install scripts / 56 manifests = very large unreviewed surface Heuristic signals Popular: 57,549 stars Licensed (MIT) Active (updated 0d ago) 6578 forks README uses curl\|bash style install Dependency manifests: 56 Contains install/shell scripts (179) Has CI workflows Has SECURITY.md
shanraisshan/claude-code-best-practiceNEW HooksMCP ServersSkillsSlash CommandsSubagents	56105	5631	HTML	MIT	2026-06-02	✦ Lower risk
from vibe coding to agentic engineering - practice makes claude perfect A best-practices collection for Claude Code (agents, commands, skills) plus an HTML presentation explaining 'vibe coding to agentic engineering'. agentic-aiagentic-codingagentic-engineeringagentic-workflowaiai-agentsanthropicbest-practicesborisclaude ✦ Claude review — Lower risk Mostly documentation and presentation skills. One skill (agent-browser) drives browser automation and a time-skill runs a harmless `date` command; both are conventional and scoped with allowed-tools. No injection or exfiltration. LOW Includes a browser-automation skill (agent-browser) scoped to its own CLI Heuristic signals Popular: 56,105 stars Licensed (MIT) Active (updated 0d ago) 5631 forks
Lum1104/Understand-AnythingNEW HooksPluginsSkillsSubagents	50261	4095	TypeScript	MIT	2026-06-02	✦ Lower risk
Graphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more. Turns a codebase or docs into an interactive knowledge graph you can query/explore; works across several agents. antigravity-skillsbusiness-knowledgeclaude-codeclaude-skillscodebase-analysiscodexcodex-skillsdeveloper-tools-ai-agentgemini-cli-skillskarpathy-llm-wiki ✦ Claude review — Lower risk Skills are read-only analysis over a local knowledge-graph JSON and a local dashboard; clean and well-scoped. The only generic cautions are a curl\|bash install and a few shell scripts. Low risk. LOW curl\|bash install; otherwise read-only local analysis Heuristic signals Popular: 50,261 stars Licensed (MIT) Active (updated 0d ago) 4095 forks README uses curl\|bash style install Dependency manifests: 5 Contains install/shell scripts (3) Has CI workflows Has SECURITY.md
santifer/career-opsNEW PluginsSkills	48418	10050	JavaScript	MIT	2026-06-02	✦ Elevated
AI-powered job search system built on Claude Code. 14 skill modes, Go dashboard, PDF generation, batch processing. An AI job-search system on Claude Code: scans job portals, generates CVs/PDFs, tracks applications, with batch processing. ai-agentanthropicautomationcareercareeropsclaudeclaude-codecligolanginterview-prep ✦ Claude review — Elevated Functionality is benign, but the batch runner is the concern: batch/batch-runner.sh launches `claude -p` workers with --dangerously-skip-permissions, i.e. it deliberately runs an agent with permission prompts disabled over your data and the web. That's a real footgun worth understanding before use. The plugin.json otherwise scopes WebFetch domains sensibly. HIGH batch-runner.sh runs `claude -p --dangerously-skip-permissions` (permission prompts disabled) for unattended workers INFO plugin.json scopes Bash(node) and WebFetch to specific job-board domains Heuristic signals Popular: 48,418 stars Licensed (MIT) Active (updated 0d ago) 10050 forks Dependency manifests: 2 Contains install/shell scripts (1) Has CI workflows Has SECURITY.md
addyosmani/agent-skillsNEW HooksPluginsSkillsSlash CommandsSubagents	47836	5294	Shell	MIT	2026-06-02	✦ Lower risk
Production-grade engineering skills for AI coding agents. Addy Osmani's production-grade engineering skills mapped to a dev lifecycle (spec, plan, build, test, review, simplify, ship). agent-skillsantigravityantigravity-ideclaude-codecursorskills ✦ Claude review — Lower risk High-quality instructional skills from a well-known author; content is process guidance, not executable payloads. Browser-testing skill relies on the Chrome DevTools MCP (declared). Installs via the official Claude plugin marketplace. Low risk. INFO Marketplace install; skills are instructional with declared MCP/tool dependencies Heuristic signals Popular: 47,836 stars Licensed (MIT) Active (updated 0d ago) 5294 forks Contains install/shell scripts (8) Has CI workflows
CherryHQ/cherry-studioNEW HooksMCP ServersPluginsResource ListsSkillsSubagents	46787	4439	TypeScript	AGPL-3.0	2026-06-03	✦ Lower risk
AI productivity studio with smart chat, autonomous agents, and 300+ assistants. Unified access to frontier LLMs An established AGPL desktop AI studio (Electron) with chat, agents, and 300+ assistants across many LLM providers. agent-skillsai-agentawesome-skillsclaude-codecodexdeepseekhermes-agentopenclawskillsvibe-coding ✦ Claude review — Lower risk Mature, widely used project. The bundled skills are repo-maintenance helpers (PR testing, gh-create-pr, code review) scoped to its own workflow; they shell out to gh/pnpm as expected for a dev repo. No injection seen. Standard desktop-app trust applies. INFO Maintenance skills shell out to gh/pnpm as normal for a dev repo Heuristic signals Popular: 46,787 stars Licensed (AGPL-3.0) Active (updated 0d ago) 4439 forks Dependency manifests: 8 Contains install/shell scripts (10) Has CI workflows Has SECURITY.md
jeecgboot/JeecgBootNEW HooksMCP ServersPluginsResource Lists	46585	16034	Java	Apache-2.0	2026-05-23	✦ Lower risk
AI 低代码平台「低代码 + 零代码」双驱动！低代码可一键生成前后端代码;零代码可 5 分钟搭建系统;AI Skills 一句话画流程、设计表单、生成整套系统。内置 AI聊天、知识库、流程编排、MCP插件等，兼容主流大模型。引领「AI 生成 → 在线配置 → 代码生成 → 手工合并->AI修改」开发模式，消除 Java 项目 80% 的重复工作，提效而不失灵活。 A long-established Chinese low-code platform (Java + Vue3) that has added AI 'Skills' for natural-language system generation. activitiagentaiantdclaude-codeclicodegeneratorcodexflowablelangchain4j ✦ Claude review — Lower risk Big, mature application framework; sampled files are ordinary Vue form hooks. The AI-skills angle is a newer add-on. Risk is the general one of adopting a large full-stack platform, not anything specific to the agent skills. Apache-2.0. INFO Mature full-stack platform; sampled code is conventional application code Heuristic signals Popular: 46,585 stars Licensed (Apache-2.0) Active (updated 10d ago) 16034 forks Dependency manifests: 31 Contains install/shell scripts (4)
hesreallyhim/awesome-claude-codeNEW HooksMCP ServersPluginsResource ListsSlash Commands	45518	3956	Python	NOASSERTION	2026-04-27	✦ Lower risk
A curated list of awesome skills, hooks, slash-commands, agent orchestrators, applications, and plugins for Claude Code by Anthropic A well-known curated list of Claude Code skills, hooks, commands, and plugins (currently mid-reorganization). agent-skillsagentic-codeagentic-codingai-workflow-optimizationai-workflowsanthropicanthropic-claudeawesomeawesome-claude-codeawesome-list ✦ Claude review — Lower risk An index repo — generated list/badge files, no meaningful executable surface. As with any awesome list, the risk is in the third-party items it links, which you vet individually. License is NOASSERTION. LOW Curated index; vet linked third-party resources individually INFO Table of contents is mid-rewrite Heuristic signals Popular: 45,518 stars No clear license Updated 36d ago 3956 forks Dependency manifests: 1 Has CI workflows Has SECURITY.md
zhayujie/CowAgentNEW MCP ServersPluginsResource ListsSkillsSlash Commands	45032	10165	Python	MIT	2026-06-02	✦ Moderate
Open-source super AI assistant & Agent Harness. Plans tasks, runs tools and skills, autonomously grows with memory and knowledge. Multi-model, multi-channel. Lightweight, extensible, one-line install. An open-source 'super assistant' agent harness that plans tasks, controls your computer and external services, and runs 24/7 across chat channels. aiai-agentai-agentschatgpt-on-wechatclaudeclaude-codecodexcowagentdeepseekharness ✦ Claude review — Moderate Capable and legitimate (from the chatgpt-on-wechat lineage), but broad by design: it controls the computer, installs Playwright/Chromium, runs as a long-lived service, and image/other skills pull in many provider API keys. One-line install. Nothing malicious in sampled code, but a 24/7 agent with system control deserves a sandbox and scoped credentials. MED Designed to control the computer and run continuously as a service; one-line install LOW Installs Playwright/Chromium and expects multiple provider API keys Heuristic signals Popular: 45,032 stars Licensed (MIT) Active (updated 0d ago) 10165 forks Dependency manifests: 2 Contains install/shell scripts (14) Has CI workflows
sickn33/antigravity-awesome-skillsNEW HooksMCP ServersPluginsResource ListsSkillsSubagents	39517	6410	Python	MIT	2026-06-02	✦ Moderate
Installable GitHub library of 1,494+ agentic skills for Claude Code, Cursor, Codex CLI, Gemini CLI, Antigravity, and more. Includes specialized plugins, installer CLI, bundles, workflows, and official/community skill collections. A very large installable library (1,490+ skills) aggregating official and community SKILL.md playbooks across many agents, with an npm installer. agent-skillsagentic-skillsai-agent-skillsai-agentsai-codingai-workflowsantigravityantigravity-skillsclaude-codeclaude-code-skills ✦ Claude review — Moderate Useful catalog, but it bundles community skills of mixed provenance — the frontmatter itself labels some skills risk: 'critical' or 'unknown', and sources vary. With 137 install scripts and 85 manifests, installing broadly pulls in a lot of third-party instruction you haven't read. Prefer installing specific vetted skills over the whole library. MED Bundles 1,490+ community skills of varying trust; some self-labeled risk critical/unknown MED 137 install scripts / 85 manifests; install narrowly, not wholesale Heuristic signals Popular: 39,517 stars Licensed (MIT) Active (updated 0d ago) 6410 forks Dependency manifests: 85 Contains install/shell scripts (137) Has CI workflows Has SECURITY.md
colbymchenry/codegraphNEW HooksMCP ServersSkills	38189	2365	TypeScript	MIT	2026-06-03	✦ Moderate
Pre-indexed code knowledge graph for Claude Code, Codex, Gemini, Cursor, OpenCode, AntiGravity, Kiro, and Hermes Agent — fewer tokens, fewer tool calls, 100% local A pre-indexed, fully-local code knowledge graph for coding agents to cut tokens and tool calls. ✦ Claude review — Moderate Local-first and genuinely useful; tests and extraction code look clean. It installs opt-in git sync hooks and ships an experimental PreToolUse hook that blocks Read of source files — fine, but one example hook-settings.json hardcodes the author's absolute path (/Users/colby/...), so don't wire that file in verbatim. curl\|bash install and 36 shell scripts. MED Installs git sync hooks and an experimental PreToolUse hook; curl\|bash install LOW Example hook-settings.json hardcodes the author's absolute path; don't use as-is Heuristic signals Popular: 38,189 stars Licensed (MIT) Active (updated 0d ago) 2365 forks README uses curl\|bash style install Dependency manifests: 2 Contains install/shell scripts (36) Has CI workflows
danny-avila/LibreChatNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	37960	7811	TypeScript	MIT	2026-06-03	✦ Lower risk
Enhanced ChatGPT Clone: Features Agents, MCP, DeepSeek, Anthropic, AWS, OpenAI, Responses API, Azure, Groq, o1, GPT-5, Mistral, OpenRouter, Vertex AI, Gemini, Artifacts, AI model switching, message search, Code Interpreter, langchain, DALL-E-3, OpenAPI Actions, Functions, Secure Multi-User Auth, Presets, open-source for self-hosting. Active. A mature, MIT-licensed self-hostable ChatGPT-style web app with agents, MCP, and multi-provider support. aianthropicartifactsawsazurechatgptchatgpt-cloneclaudeclonedeepseek ✦ Claude review — Lower risk Well-established project; sampled code is ordinary React/hooks and tests. Self-hosting it means standard web-app responsibilities (auth, keys), but nothing in the agent surface looks risky. INFO Mature web app; standard self-hosting/auth responsibilities apply Heuristic signals Popular: 37,960 stars Licensed (MIT) Active (updated 0d ago) 7811 forks Dependency manifests: 8 Contains install/shell scripts (15) Has CI workflows Has SECURITY.md
wshobson/agentsNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	36282	3933	Python	MIT	2026-06-02	✦ Lower risk
Multi-harness agentic plugin marketplace for Claude Code, Codex CLI, Cursor, OpenCode, and Gemini CLI A large 'one source, five harnesses' marketplace of agents/skills/commands (accessibility, code review, etc.). agent-skillsagentic-aiagentsai-agentsanthropicautomationclaude-codeclaude-code-pluginscodex-clicopilot ✦ Claude review — Lower risk Sampled skills are clean instructional content. Risk is breadth (84 plugins, 156 skills) rather than anything malicious; install only what you need. LOW Large catalog; install selectively rather than wholesale Heuristic signals Popular: 36,282 stars Licensed (MIT) Active (updated 0d ago) 3933 forks Dependency manifests: 2 Contains install/shell scripts (7) Has CI workflows
router-for-me/CLIProxyAPINEW HooksPlugins	35832	5966	Go	MIT	2026-06-03	✦ Moderate
Wrap Gemini CLI, Antigravity, ChatGPT Codex, Claude Code, Grok Build as an OpenAI/Gemini/Claude/Codex compatible API service, allowing you to enjoy the free Gemini 3.1 Pro, GPT 5.5, Grok 4.3, Claude model through API A proxy server that wraps CLI tools (Claude Code, Codex, Gemini, Grok) behind OpenAI/Claude-compatible APIs via OAuth and multi-account access. antigravityclaude-codecluadecodexgeminiopenai ✦ Claude review — Moderate Legitimate Go proxy, but the explicit selling point is 'free' model access through multi-account/OAuth pooling, which can run against providers' terms of service and means the tool holds OAuth credentials for multiple accounts. Use with eyes open about ToS and credential exposure. MED Aggregates multi-account OAuth credentials; 'free access' framing may violate provider ToS LOW Affiliate/relay sponsor links in README Heuristic signals Popular: 35,832 stars Licensed (MIT) Active (updated 0d ago) 5966 forks Dependency manifests: 1 Contains install/shell scripts (2) Has CI workflows
Yeachan-Heo/oh-my-claudecodeNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	35628	3248	TypeScript	MIT	2026-06-02	✦ Moderate
Teams-first Multi-agent orchestration for Claude Code A teams-first multi-agent orchestration layer for Claude Code with an 'ask' skill that routes prompts to local Claude/Codex/Gemini CLIs. agentic-codingai-agentsautomationclaudeclaude-codemulti-agent-systemsoh-my-opencodeopencodeparallel-executionvibe-coding ✦ Claude review — Moderate Functional and legit; the orchestration skills are clean. Notable surface: 167 install/shell scripts and skills that shell out to other CLIs. Nothing malicious seen, but a big install-time footprint to trust. MED 167 install/shell scripts; skills invoke external CLIs Heuristic signals Popular: 35,628 stars Licensed (MIT) Active (updated 1d ago) 3248 forks Dependency manifests: 3 Contains install/shell scripts (167) Has CI workflows Has SECURITY.md
luongnv89/claude-howtoNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	34834	4243	Python	MIT	2026-06-02	✦ Lower risk
A visual, example-driven guide to Claude Code — from basic concepts to advanced agents, with copy-paste templates that bring immediate value. A visual, example-driven Claude Code tutorial with quizzes, templates, and skills. claude-codeguidetutorial ✦ Claude review — Lower risk Educational content (lessons, quizzes, blog-draft skill). No injection or risky execution in sampled skills; 41 scripts are mostly tutorial helpers. INFO Educational/tutorial content Heuristic signals Popular: 34,834 stars Licensed (MIT) Active (updated 0d ago) 4243 forks Dependency manifests: 2 Contains install/shell scripts (41) Has CI workflows Has SECURITY.md
musistudio/claude-code-routerNEW HooksPluginsSlash CommandsSubagents	34648	2824	TypeScript	MIT	2026-03-04	✦ Moderate
Use Claude Code as the foundation for coding infrastructure, allowing you to decide how to interact with the model while enjoying updates from Anthropic. Routes Claude Code requests to other model providers, with a plugin system including a webhook output handler. ✦ Claude review — Moderate Useful infra; sampled code (webhook handler) is clean. It sits in your request path and can forward outputs to HTTP endpoints, so misconfiguration could send data off-box. Last updated ~90 days ago. MED Proxies requests and can POST outputs to webhooks; review routing/output config LOW curl\|bash install; ~90 days since last update Heuristic signals Popular: 34,648 stars Licensed (MIT) Updated 90d ago 2824 forks README uses curl\|bash style install Dependency manifests: 7 Contains install/shell scripts (8) Has CI workflows
Alishahryar1/free-claude-codeNEW Subagents	31884	4836	Python	MIT	2026-06-02	✦ Moderate
Use claude-code for free in the terminal, VSCode extension or discord like OpenClaw (voice supported) Tooling to use Claude Code 'for free' through your own Anthropic-compatible proxy, across terminal/VS Code/chat bots. ✦ Claude review — Moderate Installer (install.sh/.ps1) bootstraps uv, Python 3.14, and Claude Code itself — a lot happening at install time, and the 'free' premise leans on proxying to alternative providers (ToS-adjacent). Scripts looked conventional, but it's a broad install to trust. MED Install scripts bootstrap a full toolchain (uv/Python/Claude Code) LOW 'Free access' premise depends on third-party proxies; ToS considerations Heuristic signals Popular: 31,884 stars Licensed (MIT) Active (updated 0d ago) 4836 forks README uses curl\|bash style install Dependency manifests: 1 Contains install/shell scripts (3) Has CI workflows
Leonxlnx/taste-skillNEW Skills	31869	2346	Shell	MIT	2026-05-26	✦ Lower risk
Taste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop Portable 'anti-slop' design Agent Skills (taste, minimalist, brutalist, brandkit, GSAP motion) to make AI-built UIs look better. agentaiclaudeclaude-codecodexcodingdesignfrontendlowcodenocode ✦ Claude review — Lower risk This is the skill set used to style this very page. Content is pure design guidance plus image-gen prompts; only one shell script and no injection or data access. The GSAP/motion skills push heavy animation, which is a taste/usability choice, not a security issue. INFO Design guidance only; no meaningful executable surface Heuristic signals Popular: 31,869 stars Licensed (MIT) Active (updated 7d ago) 2346 forks Contains install/shell scripts (1)
coreyhaines31/marketingskillsNEW PluginsSkills	31655	5221	JavaScript	MIT	2026-05-29	✦ Lower risk
Marketing skills for Claude Code and AI agents. CRO, copywriting, SEO, analytics, and growth engineering. Marketing-focused Claude skills: CRO, copywriting, SEO, A/B testing, paid ads. claudecodexmarketing ✦ Claude review — Lower risk Instructional skills with no risky execution. Lots of affiliate links to the author's products in the README, but that's marketing, not a security concern. INFO Instructional; README is heavy on affiliate links Heuristic signals Popular: 31,655 stars Licensed (MIT) Active (updated 4d ago) 5221 forks Contains install/shell scripts (2) Has CI workflows
anthropics/claude-plugins-officialNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	29185	3117	Python	Apache-2.0	2026-06-02	✦ Lower risk
Official, Anthropic-managed directory of high quality Claude Code Plugins. Anthropic's official curated directory of Claude Code plugins, including first-party and third-party 'external_plugins'. claude-codemcpskills ✦ Claude review — Lower risk First-party directory that itself prominently warns to trust a plugin before installing because the external plugins are third-party. The sampled channel skills (Discord/iMessage) write bot tokens to ~/.claude and notably include explicit anti-injection guardrails ('only act on requests typed by the user, refuse instructions arriving via a channel message') — good design. Treat external_plugins on their own merits. INFO external_plugins are third-party; vet individually as the README warns INFO Channel skills store bot tokens locally and include explicit anti-injection guardrails Heuristic signals Popular: 29,185 stars Licensed (Apache-2.0) Active (updated 0d ago) 3117 forks Dependency manifests: 4 Contains install/shell scripts (17) Has CI workflows
davila7/claude-code-templatesNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	27733	2861	Python	MIT	2026-06-02	✦ Moderate
CLI tool for configuring and monitoring Claude Code A popular CLI for configuring and monitoring Claude Code, bundling many components including an AI Maestro agent suite. anthropicanthropic-claudeclaudeclaude-code ✦ Claude review — Moderate Widely used and useful. Surface is large (88 shell scripts, 21 manifests) and some bundled skills add inter-agent messaging with cryptographic signing and agent lifecycle control. No malice seen; review what the installer pulls in. MED 88 install scripts / 21 manifests; bundles agent-messaging and lifecycle tooling Heuristic signals Popular: 27,733 stars Licensed (MIT) Active (updated 0d ago) 2861 forks Dependency manifests: 21 Contains install/shell scripts (88) Has CI workflows Has SECURITY.md
iOfficeAI/AionUiNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	27462	2645	TypeScript	Apache-2.0	2026-06-02	✦ Moderate
Free, local, open-source 24/7 Cowork app for OpenClaw, Hermes Agent, Claude Code, Codex, OpenCode, Gemini CLI and 20+ more CLI \| Customize your assistants \| Star if you like it! A free local desktop 'Cowork' app that runs many CLI agents 24/7 with remote access and automation. acpagent-teamaiai-agentchatchatbotclaude-codeclawdbotcodexcowork ✦ Claude review — Moderate Capable Electron app; skills are repo-maintenance helpers, but one (fix-issues) is an autonomous daemon that fetches GitHub bug issues, edits code, and opens PRs on its own. A 24/7 agent with remote access and auto-fix deserves scoped credentials and a sandbox. MED Runs agents 24/7 with remote access; fix-issues daemon auto-edits code and opens PRs Heuristic signals Popular: 27,462 stars Licensed (Apache-2.0) Active (updated 0d ago) 2645 forks Dependency manifests: 6 Contains install/shell scripts (17) Has CI workflows
K-Dense-AI/scientific-agent-skillsNEW MCP ServersSkills	27015	2791	Python	MIT	2026-06-01	✦ Moderate
Turn any AI agent into an AI Scientist. The #1 Agent Skills library for science, used by 160,000+ scientists worldwide. 140 ready-to-use skills plus 100+ scientific databases covering biology, chemistry, medicine, and drug discovery. Compatible with Cursor, Claude Code, Codex, Antigravity, and the open Agent Skills standard. A large library of 140+ science skills (biology, chemistry, drug discovery) wrapping scientific APIs and Python SDKs. agent-skillsai-scientistbioinformaticschemoinformaticsclaudeclaude-skillsclaudecodeclinical-researchcomputational-biologydata-analysis ✦ Claude review — Moderate Legitimate and well-documented; many skills require external accounts/API keys (e.g. Adaptyv) and run Python via uv. Risk is the usual curl\|bash install plus the breadth of third-party scientific tooling it drives. The repo runs its own security-scan CI, which is a plus. MED Skills drive many external scientific APIs and run Python via uv; curl\|bash install INFO Repo runs a security-scan CI workflow Heuristic signals Popular: 27,015 stars Licensed (MIT) Active (updated 1d ago) 2791 forks README uses curl\|bash style install Dependency manifests: 1 Contains install/shell scripts (5) Has CI workflows Has SECURITY.md
mvanhorn/last30days-skillNEW HooksPluginsSkillsSubagents	27011	2314	Python	MIT	2026-06-01	✦ Moderate
AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary A research skill that pulls recent posts/engagement from Reddit, X, YouTube, TikTok, HN, Polymarket and synthesizes a summary. ai-promptsai-skillblueskyclaudeclaude-codeclawhubdeep-researchhackernewsinstagramopenclaw ✦ Claude review — Moderate Useful aggregator. Two things to note: a SessionStart hook auto-runs check-config.sh on every session, and the skill scrapes/queries many third-party platforms (and likely their APIs/keys). Benign as sampled, but the auto-run hook and external data fetching warrant awareness. MED SessionStart hook auto-runs a bash script every session LOW Queries many third-party social platforms; may need API keys Heuristic signals Popular: 27,011 stars Licensed (MIT) Active (updated 1d ago) 2314 forks Dependency manifests: 2 Contains install/shell scripts (5) Has CI workflows
Imbad0202/academic-research-skillsNEW HooksPluginsSkillsSlash CommandsSubagents	26297	2166	Python	NOASSERTION	2026-06-02	✦ Lower risk
Academic Research Skills for Claude Code: research → write → review → revise → finalize An end-to-end academic writing pipeline (research, write, integrity check, peer-review simulation, revise, finalize). academic-pipelineacademic-writingai-researchclaudeclaude-codeliterature-reviewpeer-reviewprompt-engineering ✦ Claude review — Lower risk Instructional multi-agent writing skills; no risky execution in sampled content. License is CC BY-NC (non-commercial), worth noting for reuse. Marketplace install. INFO Instructional pipeline LOW Non-commercial license (CC BY-NC); check before commercial use Heuristic signals Popular: 26,297 stars No clear license Active (updated 0d ago) 2166 forks Contains install/shell scripts (2) Has CI workflows Has SECURITY.md
jarrodwatts/claude-hudNEW PluginsSlash Commands	24339	1099	JavaScript	MIT	2026-05-29	✦ Lower risk
A Claude Code plugin that shows what's happening - context usage, active tools, running agents, and todo progress A Claude Code statusline HUD showing context usage, active tools, running agents, and todo progress. anthropicclaudeclaude-codeclipluginstatuslinetypescript ✦ Claude review — Lower risk Small, focused TypeScript plugin that reads session state to render a statusline; no network or risky execution in sampled code. Highest heuristic score of the set. INFO Read-only statusline display Heuristic signals Popular: 24,339 stars Licensed (MIT) Active (updated 4d ago) 1099 forks Dependency manifests: 1 Has CI workflows Has SECURITY.md
VoltAgent/awesome-agent-skillsNEW Resource Lists	24046	2580	—	MIT	2026-05-27	✦ Lower risk
A curated collection of 1000+ agent skills from official dev teams and the community, compatible with Claude Code, Codex, Gemini CLI, Cursor, and more. A curated, hand-picked list of 1,000+ agent skills from teams and the community. agent-skillsai-agentsantigravity-skillsawesomeawesome-listawesome-listsclaude-codeclaude-code-skillsclaude-skillscodex-skills ✦ Claude review — Lower risk An index repo (no bundled executable surface in the scan). As always with awesome lists, vet each linked skill on its own before installing. LOW Curated index; vet linked skills individually Heuristic signals Popular: 24,046 stars Licensed (MIT) Active (updated 6d ago) 2580 forks
SuperClaude-Org/SuperClaude_FrameworkNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	23145	1969	Python	MIT	2026-04-27	✦ Lower risk
A configuration framework that enhances Claude Code with specialized commands, cognitive personas, and development methodologies. A configuration framework adding structured commands, personas, and methodologies to Claude Code. ✦ Claude review — Lower risk Sampled skills (confidence-check, brainstorm) are read-only/instructional and scope allowed-tools sensibly. Mostly config and prompting; modest install footprint. Last updated ~37 days ago. INFO Config/prompting framework; skills scope tools explicitly Heuristic signals Popular: 23,145 stars Licensed (MIT) Updated 37d ago 1969 forks Dependency manifests: 2 Contains install/shell scripts (18) Has CI workflows Has SECURITY.md
OthmanAdi/planning-with-filesNEW HooksPluginsSkillsSlash Commands	22578	2001	Python	MIT	2026-05-26	✦ Moderate
Claude Code skill implementing Manus-style persistent markdown planning — the workflow pattern behind the $2B acquisition. A persistent markdown planning skill (task_plan/findings/progress) modeled on Manus-style working memory. adalagent-skillsantigravityclaudeclaude-codeclaude-skillscopilotcopilot-skillshermeshermes-agent ✦ Claude review — Moderate Genuinely useful pattern and the repo advertises a security audit. The notable behavior: a UserPromptSubmit hook embeds shell that runs on every prompt (validating a PLAN_ID slug), and there are 149 install scripts. The slug is regex-validated, which is reassuring, but an auto-run-on-every-prompt hook plus that many scripts is worth understanding before install. MED UserPromptSubmit hook runs embedded shell on every prompt (PLAN_ID is regex-validated) LOW 149 install scripts Heuristic signals Popular: 22,578 stars Licensed (MIT) Active (updated 7d ago) 2001 forks Dependency manifests: 1 Contains install/shell scripts (149) Has CI workflows
blader/humanizerNEW Skills	22125	2113	—	MIT	2026-05-27	✦ Lower risk
Claude Code skill that removes signs of AI-generated writing from text A single skill that strips tell-tale signs of AI writing to make text read more naturally. ✦ Claude review — Lower risk One SKILL.md of editing guidance, scoped to read/write/edit on text. Install is a plain git clone into the skills directory. No risky surface. INFO Single text-editing skill; git-clone install Heuristic signals Popular: 22,125 stars Licensed (MIT) Active (updated 7d ago) 2113 forks
VoltAgent/awesome-claude-code-subagentsNEW MCP ServersPluginsResource ListsSubagents	21097	2469	Shell	MIT	2026-05-27	✦ Lower risk
A collection of 100+ specialized Claude Code subagents covering a wide range of development use cases A curated collection of 150+ specialized Claude Code subagents grouped into plugin categories. ai-agent-frameworkai-agent-toolsai-agentsawesomeawesome-listclaudeclaude-aiclaude-code-subagentsclaude-subagentssubagents ✦ Claude review — Lower risk Subagent definitions (markdown personas) packaged as plugins; no risky execution in sampled manifests. Vet individual agents you enable, but low risk overall. LOW Many subagent definitions; review the ones you enable Heuristic signals Popular: 21,097 stars Licensed (MIT) Active (updated 6d ago) 2469 forks Contains install/shell scripts (3) Has CI workflows
Donchitos/Claude-Code-Game-StudiosNEW HooksPluginsSkillsSubagents	20672	3016	Shell	MIT	2026-05-21	✦ Lower risk
Turn Claude Code into a full game dev studio — 49 AI agents, 72 workflow skills, and a complete coordination system mirroring real studio hierarchy. A game-dev 'studio' setup: 49 agents and 70+ skills mirroring a real studio hierarchy (ADRs, architecture review, etc.). ai-agentsai-assisted-developmentanthropicclaudeclaude-codegame-designgame-developmentgamedevgodotindie-game-dev ✦ Claude review — Lower risk Sampled skills are well-scoped planning/review workflows (Read/Glob/Grep/Write). 13 scripts and a dozen hooks, but nothing risky in the sampled content. Low-to-moderate by virtue of size. INFO Large but well-scoped planning/review skills Heuristic signals Popular: 20,672 stars Licensed (MIT) Active (updated 12d ago) 3016 forks Contains install/shell scripts (13) Has SECURITY.md
openai/codex-plugin-ccNEW HooksPluginsSkillsSlash CommandsSubagents	20158	1220	JavaScript	Apache-2.0	2026-04-18	✦ Lower risk
Use Codex from Claude Code to review code or delegate tasks. OpenAI's official plugin to invoke Codex from inside Claude Code for reviews or delegated tasks. ✦ Claude review — Lower risk First-party (OpenAI), Apache-licensed. Internal helper skills are tightly scoped forwarders to a companion script with explicit 'do one task and return stdout unchanged' rules. Sends code to Codex/OpenAI by design (usage/limits apply). Clean. INFO Delegates code to Codex/OpenAI by design; first-party and tightly scoped Heuristic signals Popular: 20,158 stars Licensed (Apache-2.0) Updated 45d ago 1220 forks Dependency manifests: 1 Has CI workflows
zarazhangrui/frontend-slidesNEW PluginsSkills	20068	1644	JavaScript	MIT	2026-05-26	✦ Lower risk
Create beautiful slides on the web using a coding agent's frontend skills A skill for generating zero-dependency, single-file HTML presentations (and converting PPTX), with an anti-slop design philosophy. ai-slidesanthropicclaudeclaude-codeclaude-skillgenerative-uihtmlpresentationslidesvibe-coding ✦ Claude review — Lower risk Self-contained HTML output, no build tools or network surface in the skill itself. A few shell scripts for PPT conversion. Low risk. INFO Produces single-file HTML; minimal surface Heuristic signals Popular: 20,068 stars Licensed (MIT) Active (updated 7d ago) 1644 forks Contains install/shell scripts (4)
EveryInc/compound-engineering-pluginNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	19426	1444	TypeScript	MIT	2026-06-03	✦ Lower risk
Official Compound Engineering plugin for Claude Code, Codex, Cursor, and more Every Inc's official 'compound engineering' plugin: planning/review-heavy skills and agents plus a coding tutor. compoundengineering ✦ Claude review — Lower risk Instructional planning/review/architecture-audit skills; sampled content is clean and process-oriented. Active, MIT, has CI and SECURITY.md. Low risk. INFO Planning/review-focused instructional skills Heuristic signals Popular: 19,426 stars Licensed (MIT) Active (updated 0d ago) 1444 forks Dependency manifests: 2 Contains install/shell scripts (18) Has CI workflows Has SECURITY.md
alirezarezvani/claude-skillsNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	16962	2326	Python	MIT	2026-06-02	✦ Moderate
337 Claude Code skills & agent skills & plugins (30+ Agents, 70+ custom commands, 330+ skills, customizable references, scripts)for Claude Code, Codex, Gemini CLI, Cursor, and 8 more coding agents — engineering, marketing, product, compliance, C-level advisory, research, business operations, commercial & finance, and your daily productivity skills. A 330+ skill/agent library spanning engineering, marketing, compliance, C-level personas, and research, for many coding agents. agent-pluginsagent-skillsagentic-aiai-coding-agentanthropic-claudeclaude-aiclaude-codeclaude-code-pluginsclaude-code-skillsclaude-skills ✦ Claude review — Moderate Broad, actively maintained collection; sampled persona/skill templates are clean instructional content and it advertises PreToolUse security hooks. Risk is breadth — installing everything pulls in a lot of mixed-purpose personas and 14 scripts. Install selectively. MED Very large mixed skill/persona bundle; install selectively Heuristic signals Popular: 16,962 stars Licensed (MIT) Active (updated 0d ago) 2326 forks Dependency manifests: 5 Contains install/shell scripts (14) Has CI workflows Has SECURITY.md
mksglu/context-modeNEW HooksMCP ServersPluginsSkills	16293	1174	TypeScript	NOASSERTION	2026-06-02	✦ Moderate
Context window optimization for AI coding agents. Sandboxes tool output, 98% reduction. 15 platforms Context-window optimization that sandboxes tool output via an MCP (ctx_execute) to cut tokens ~98%. antigravityclaudeclaude-codeclaude-code-hooksclaude-code-pluginsclaude-code-skillcodexcodex-clicontext-modecopilot ✦ Claude review — Moderate The end-user skill is a sensible MCP-based output sandbox. One bundled skill (context-mode-ops, the owner's own ops tooling) opens with an 'ABSOLUTE, NON-NEGOTIABLE ... supersedes all other sections' override preamble — that authority-claiming framing is exactly the pattern to be wary of in a skill, though here it targets the maintainer's repo ops rather than end users. No clear license. MED context-mode-ops skill uses an 'absolute, supersedes-all' override preamble (authority-claiming framing) LOW No clear license Heuristic signals Popular: 16,293 stars No clear license Active (updated 0d ago) 1174 forks Dependency manifests: 4 Contains install/shell scripts (10) Has CI workflows
JCodesMore/ai-website-cloner-templateNEW HooksSkillsSlash Commands	16087	2456	TypeScript	MIT	2026-06-01	✦ Moderate
Clone any website with one command using AI coding agents A template that 'clones' any website from a URL — extracts assets/CSS/content and dispatches parallel builder agents to rebuild it in Next.js. aiai-agentsai-toolsautomationboilerplateclaudeclaude-codeclonedeveloper-toolsnextjs ✦ Claude review — Moderate Technically clean (well-scoped skill, one script), but the core use is copying other people's sites wholesale, which raises IP/copyright and ToS questions, and it autonomously spawns parallel builders in worktrees. Use only on sites you own or have rights to. MED Clones arbitrary third-party sites (IP/copyright/ToS considerations) LOW Auto-dispatches parallel builder agents in git worktrees Heuristic signals Popular: 16,087 stars Licensed (MIT) Active (updated 1d ago) 2456 forks Dependency manifests: 1 Contains install/shell scripts (1) Has CI workflows
wasp-lang/open-saasNEW HooksSkills	14605	1737	TypeScript	MIT	2026-06-01	✦ Lower risk
A 100% free modern JS SaaS boilerplate (React, NodeJS, Prisma). Full-featured: Auth (email, google, github, slack, MS), Email sending, Background jobs, Landing page, Payments (Stripe, Polar.sh), Shadcn UI, S3 file upload. AI-ready with tailored AGENTS.md, skills, and Claude Code plugin. One cmd deploy. Powered by Wasp full-stack framework. A free, MIT SaaS boilerplate (React/Node/Prisma) with auth, payments, and AI-ready skills/AGENTS.md. aiauthenticationaws-s3boilerplatechatgptfull-stackgoogle-authhacktoberfestnodejsopen-source ✦ Claude review — Lower risk Established starter kit; sampled skills mainly fetch the project's own docs from raw.githubusercontent and guide setup. Standard boilerplate responsibilities; nothing risky in the agent surface. INFO Skills fetch the project's own docs remotely; standard boilerplate otherwise Heuristic signals Popular: 14,605 stars Licensed (MIT) Active (updated 1d ago) 1737 forks Dependency manifests: 5 Contains install/shell scripts (5) Has CI workflows
travisvn/awesome-claude-skillsNEW Resource Lists	13123	1455	—	None	2026-04-28	✦ Lower risk
A curated list of awesome Claude Skills, resources, and tools for customizing Claude AI workflows — particularly Claude Code A curated list of Claude Skills, resources, and tools. agentic-codinganthropicawesomeawesome-listawesome-listsclaudeclaude-aiclaude-codeclaude-desktopclaude-skills ✦ Claude review — Lower risk Index repo with no bundled executable surface. Vet linked skills individually. No repo license set. LOW Curated index; no license; vet linked items Heuristic signals Popular: 13,123 stars No clear license Updated 35d ago 1455 forks
YishenTu/claudianNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	12227	748	TypeScript	MIT	2026-06-01	✦ Moderate
An Obsidian plugin that embeds Claude Code/Codex as an AI collaborator in your vault An Obsidian plugin that embeds Claude Code/Codex as an agent inside your vault with file read/write, search, and bash. claude-codecodexideobsidianobsidian-pluginproductivity ✦ Claude review — Moderate Useful and the sampled hook code (stop-subagent guard) is clean. By design it gives an agent file and bash access scoped to your vault, and package.json runs a postinstall script. Reasonable, but understand it's an agent with shell access to your notes directory. MED Grants an agent file/bash access within the Obsidian vault; npm postinstall script Heuristic signals Popular: 12,227 stars Licensed (MIT) Active (updated 1d ago) 748 forks Dependency manifests: 1 Contains install/shell scripts (1) Has CI workflows
BeehiveInnovations/pal-mcp-serverNEW MCP ServersSlash CommandsSubagents	11578	1010	Python	NOASSERTION	2025-12-15	✦ Lower risk
The power of Claude Code / GeminiCLI / CodexCLI + [Gemini / OpenAI / OpenRouter / Azure / Grok / Ollama / Custom Model / All Of The Above] working as one. A provider-abstraction MCP (formerly Zen MCP) that lets one CLI use many models in a single workflow, with a CLI-to-CLI bridge. ✦ Claude review — Lower risk Mature multi-model MCP; sampled scripts are ordinary lint/build/quality checks. It routes prompts to multiple providers (keys/config apply) and is ~169 days stale with no clear license, but no risky patterns seen. LOW Routes prompts to multiple model providers; ~169 days since update; no clear license Heuristic signals Popular: 11,578 stars No clear license Updated 169d ago 1010 forks Dependency manifests: 2 Contains install/shell scripts (10) Has CI workflows Has SECURITY.md
can1357/oh-my-piNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	10035	831	TypeScript	MIT	2026-06-03	✦ Moderate
⌥ AI Coding agent for the terminal — hash-anchored edits, optimized tool harness, LSP, Python, browser, subagents, and more A full terminal coding agent ('pi') with IDE/LSP wiring, Python, browser, and subagents. ai-agentai-coding-agentanthropicbunclaudeclicoding-assistantllmmcpmulti-provider ✦ Claude review — Moderate It's a complete coding-agent harness, so by nature it edits files and runs commands; sampled skills (semantic-compression, system-prompts) are clean prompt-engineering guidance. Risk is the broad capability surface plus a curl\|bash install and 26 dependency manifests — adopt it as you would any autonomous coding agent. MED Full autonomous coding agent (file edits, shell, browser); curl\|bash install Heuristic signals Popular: 10,035 stars Licensed (MIT) Active (updated 0d ago) 831 forks README uses curl\|bash style install Dependency manifests: 26 Contains install/shell scripts (23) Has CI workflows Has SECURITY.md
diet103/claude-code-infrastructure-showcaseNEW HooksSkillsSlash CommandsSubagents	9692	1221	Shell	MIT	2026-04-17	✦ Lower risk
Examples of my Claude Code infrastructure with skill auto-activation, hooks, and agents A reference library of Claude Code infrastructure patterns (auto-activating skills via hooks, agents, dev-docs). ✦ Claude review — Lower risk Explicitly a copy-what-you-need reference, not a runnable app. Sampled skills are dev guidelines. Hooks auto-activate skills; review those before wiring them in, but low risk overall. INFO Reference patterns to copy; hooks auto-activate skills Heuristic signals Popular: 9,692 stars Licensed (MIT) Updated 46d ago 1221 forks Dependency manifests: 1 Contains install/shell scripts (6)
ykdojo/claude-code-tipsNEW HooksMCP ServersPluginsSkillsSlash Commands	8594	651	JavaScript	NOASSERTION	2026-05-06	✦ Moderate
45 tips for getting the most out of Claude Code, from basics to advanced - includes a custom status line script, cutting the system prompt in half, using Gemini CLI as Claude Code's minion, and Claude Code running itself in a container. Also includes the dx plugin. A 45-tip Claude Code guide plus a 'dx' plugin, including conversation-cloning scripts and running Claude in a container. agenticagentic-aiagentic-codingagentic-workflowaiclaudeclaude-aiclaude-codeclideveloper-tools ✦ Claude review — Moderate Mostly tips, but it ships ~149 scripts and skills that read your session history (~/.claude/history.jsonl) and run helper scripts to clone/trim conversations. Benign as sampled, but the lowest heuristic score here, no clear license, and a lot of session-manipulating scripting to trust. MED Skills read ~/.claude session history and run many helper scripts; no clear license Heuristic signals Popular: 8,594 stars No clear license Active (updated 27d ago) 651 forks README uses curl\|bash style install Dependency manifests: 2 Contains install/shell scripts (149)
idosal/git-mcpNEW HooksMCP ServersPlugins	8124	719	TypeScript	Apache-2.0	2026-05-08	✦ Lower risk
Put an end to code hallucinations! GitMCP is a free, open-source, remote MCP server for any GitHub project A free, open-source remote MCP server that exposes any GitHub project to reduce code hallucinations. agentic-aiagentsaiclaudecopilotcursorgitllmmcp ✦ Claude review — Lower risk Clean Apache-licensed web app; sampled code is ordinary React hooks. It's a hosted/remote MCP, so be aware your queries go to the GitMCP service, but the project itself is sound and high-scoring. INFO Hosted remote MCP; queries go to the GitMCP service Heuristic signals Popular: 8,124 stars Licensed (Apache-2.0) Active (updated 25d ago) 719 forks Dependency manifests: 1 Has CI workflows Has SECURITY.md
dontriskit/awesome-ai-system-promptsNEW Resource Lists	5942	891	TypeScript	MIT	2026-02-20	✦ Lower risk
🧠 Curated collection of system prompts for top AI tools. Perfect for AI agent builders and prompt engineers. Incuding: ChatGPT, Claude, Perplexity, Manus, Claude-Code, Loveable, v0, Grok, same new, windsurf, notion, and MetaAI. A curated collection of system prompts from many AI tools, framed as prompt-engineering guidance. ✦ Claude review — Lower risk Reference text only, no executable surface. Provenance of some prompts is uncertain, but no technical risk. ~102 days stale. INFO Reference prompts only; no executable surface Heuristic signals Popular: 5,942 stars Licensed (MIT) Updated 102d ago 891 forks
ChrisWiles/claude-code-showcaseNEW HooksMCP ServersSkillsSlash CommandsSubagents	5939	563	JavaScript	None	2026-01-06	✦ Lower risk
Comprehensive Claude Code project configuration example with hooks, skills, agents, commands, and GitHub Actions workflows An example Claude Code project configuration: skills, agents, commands, hooks, and CI workflows. ✦ Claude review — Lower risk Sampled skills are coding-convention guides (components, Formik, GraphQL). Reference configuration; no risky execution. No clear license and ~147 days stale. LOW Example config; no clear license; somewhat stale Heuristic signals Popular: 5,939 stars No clear license Updated 147d ago 563 forks Contains install/shell scripts (1) Has CI workflows
google-labs-code/stitch-skillsNEW MCP ServersPluginsSkills	5860	716	TypeScript	Apache-2.0	2026-06-02	✦ Lower risk
A library of Agent Skills designed to work with the Stitch MCP server. Each skill follows the Agent Skills open standard, for compatibility with coding agents such as Antigravity, Gemini CLI, Claude Code, Cursor. Google's official Agent Skills for the Stitch design MCP (design-to-React, Remotion videos, shadcn/ui). ✦ Claude review — Lower risk First-party (Google Labs), Apache-licensed; skills scope allowed-tools to the Stitch MCP namespace plus standard file/bash/web_fetch. Clean and well-structured. Uses the Stitch MCP service by design. INFO First-party; scopes tools to the Stitch MCP; uses the Stitch service Heuristic signals Popular: 5,860 stars Licensed (Apache-2.0) Active (updated 0d ago) 716 forks Dependency manifests: 2 Contains install/shell scripts (4) Has CI workflows Has SECURITY.md
entireio/cliNEW HooksPluginsSkillsSlash CommandsSubagents	4457	342	Go	MIT	2026-06-03	✦ Moderate
📜 Entire CLI hooks into your Git workflow to capture AI agent sessions as you work. Sessions are indexed alongside commits, creating a searchable record of how code was written in your repo. A CLI that hooks into Git to capture AI agent sessions (prompts, transcripts, tool calls, token usage) indexed alongside commits. agentsaiclaudedeveloperdeveloper-platformgemini ✦ Claude review — Moderate Useful for traceability and the sampled skills are clean pipelines. By design it records full prompt/response transcripts and stores them on a branch, so it's a data-capture tool — fine for audit use, but know that complete session content is being persisted. curl\|bash install. MED By design records full agent transcripts/prompts alongside commits LOW curl\|bash install Heuristic signals Popular: 4,457 stars Licensed (MIT) Active (updated 0d ago) 342 forks README uses curl\|bash style install Dependency manifests: 1 Contains install/shell scripts (16) Has CI workflows Has SECURITY.md
vijaythecoder/awesome-claude-agentsNEW Subagents	4292	521	—	MIT	2025-10-30	✦ Lower risk
An orchestrated sub agent dev team powered by claude code An orchestrated team of specialized Claude Code subagents for full-stack development. ✦ Claude review — Lower risk Subagent definitions installed via git clone/symlink; the README is upfront that it's experimental and token-intensive. No risky execution in the surface; low risk. INFO Experimental, token-intensive; subagent definitions via git clone Heuristic signals Popular: 4,292 stars Licensed (MIT) 521 forks
zebbern/claude-code-guideNEW HooksMCP ServersSkillsSubagents	4228	424	Python	MIT	2026-06-03	✦ Elevated
Claude Code Guide - Setup, Commands, workflows, agents, skills & tips-n-tricks go from beginner to power user! A Claude Code guide that also bundles a set of offensive-security skills (Active Directory attacks, Kerberoasting, API fuzzing, bug-bounty exploitation). aiai-agentai-agent-toolsanthropic-claudeclaudeclaude-aiclaude-apiclaude-codeclaude-code-communicationclaude-code-guide ✦ Claude review — Elevated Beyond the docs, this repo packages dual-use offensive tooling — skills for AD exploitation (DCSync, Golden Ticket, pass-the-hash) and API/IDOR fuzzing — that an agent can auto-load. Legitimate for authorized pentesting, but powerful and easily misused, and the scan also flagged a possible hardcoded secret in the README plus a confusing license (MIT in code, 'Anthropic' on a badge). Treat with care and only in authorized engagements. HIGH Bundles auto-loadable offensive-security skills (AD attacks, Kerberoasting, exploitation) MED Heuristic flagged a possible hardcoded secret in the README LOW Inconsistent license signals (MIT vs 'Anthropic' badge) Heuristic signals Popular: 4,228 stars Licensed (MIT) Active (updated 0d ago) 424 forks Possible hardcoded secret(s) in README (1) README uses curl\|bash style install Dependency manifests: 2 Contains install/shell scripts (5) Has CI workflows Has SECURITY.md
zhukunpenglinyutong/jetbrains-cc-guiNEW HooksMCP ServersPluginsSkillsSubagents	3844	491	TypeScript	MIT	2026-06-01	✦ Lower risk
Jetbrains Claude Code and Codex GUI Plugin An IntelliJ/JetBrains GUI plugin for driving Claude Code and Codex. ✦ Claude review — Lower risk Sampled content includes a Vercel-authored React best-practices skill and ordinary webview hooks. The maintainer states they run /security-review before releases. Clean IDE plugin. INFO IDE GUI plugin; maintainer runs security review per release Heuristic signals Popular: 3,844 stars Licensed (MIT) Active (updated 1d ago) 491 forks Dependency manifests: 2 Contains install/shell scripts (2) Has CI workflows
parcadei/Continuous-Claude-v3NEW HooksMCP ServersPluginsSkillsSubagents	3797	298	Python	MIT	2026-01-26	✦ Moderate
Context management for Claude Code. Hooks maintain state via ledgers and handoffs. MCP execution without context pollution. Agent orchestration with isolated context windows. A persistent multi-agent context-management layer for Claude Code (ledgers, handoffs, agent orchestration), 109 skills / 30 hooks. agentsclaude-codeclaude-code-cliclaude-code-hooksclaude-code-mcpclaude-code-skillsclaude-code-subagentsclaude-skillsmcp ✦ Claude review — Moderate Sensible context-isolation patterns (background agents writing to files). It bundles a Braintrust tracing plugin that sends your Claude Code conversations to an external observability service — useful but a data-egress path to understand. ~127 days stale, 50 scripts. MED Bundled tracing plugin sends conversations to external Braintrust observability LOW 50 scripts / 30 hooks; ~127 days since update Heuristic signals Popular: 3,797 stars Licensed (MIT) Updated 127d ago 298 forks README uses curl\|bash style install Dependency manifests: 2 Contains install/shell scripts (50)
disler/claude-code-hooks-masteryNEW HooksMCP ServersSlash CommandsSubagents	3733	616	Python	None	2026-03-04	✦ Moderate
Master Claude Code Hooks An educational repo demonstrating the full Claude Code hook lifecycle, sub-agents, and a meta-agent. ✦ Claude review — Moderate Good learning resource, but by topic it ships working hooks that intercept the agent: a PermissionRequest hook that can auto-allow/deny or modify tool inputs, and notification hooks that call TTS via ElevenLabs/OpenAI keys (uv-run single-file scripts). Educational and transparent, but those hooks are powerful; review before enabling. No clear license. MED Ships a PermissionRequest hook that can auto-allow/deny/modify tool inputs LOW uv-run hook scripts use TTS API keys; no clear license Heuristic signals Popular: 3,733 stars No clear license Updated 90d ago 616 forks README uses curl\|bash style install Dependency manifests: 1
glitternetwork/pinmeNEW Skills	3606	265	TypeScript	MIT	2026-05-27	✦ Moderate
Deploy Your Frontend in a Single Command. Claude Code Skills supported. A one-command frontend/full-stack deploy CLI (with web3/IPFS-style hosting) and Claude skills for auth/email/LLM via PinMe's platform. ai-toolsclaude-code-skillclaude-skillsdeploymentdeployment-toolsfrontendfrontend-deploymenthostingserverlessskills ✦ Claude review — Moderate Convenient deploy tool; skills generate Worker code that calls PinMe's proxied auth/email/OpenRouter APIs using a project API key (real provider keys stay server-side, which is good). Risk is that deploying and proxying routes your app and traffic through PinMe's platform — fine if intended, worth knowing. MED Deploys to and proxies APIs through the PinMe platform INFO Workers use a project key; real provider keys are not exposed client-side Heuristic signals Popular: 3,606 stars Licensed (MIT) Active (updated 6d ago) 265 forks Dependency manifests: 2 Contains install/shell scripts (2)
matt1398/claude-devtoolsNEW HooksSlash CommandsSubagents	3504	264	TypeScript	MIT	2026-05-13	✦ Lower risk
The missing DevTools for Claude Code — inspect session logs, tool calls, token usage, subagents, and context window in a visual UI. Free, open source. A visual DevTools app to inspect Claude Code session logs, tool calls, token usage, and context locally. aiai-agentai-debuggingai-toolsanthropicclaudeclaude-codeclaude-code-toolsdebuggingdesktop-app ✦ Claude review — Lower risk Reads local Claude Code logs to render a debugging UI; sampled code is ordinary React state/hooks. Local, read-only observability; clean and high-scoring. INFO Local read-only inspection of session logs Heuristic signals Popular: 3,504 stars Licensed (MIT) Active (updated 20d ago) 264 forks Dependency manifests: 1 Contains install/shell scripts (1) Has CI workflows Has SECURITY.md
gotalab/cc-sddNEW SkillsSlash CommandsSubagents	3429	256	TypeScript	MIT	2026-05-20	✦ Lower risk
Turn approved specs into long-running autonomous implementation. A minimal, adaptable SDD harness with Agent Skills for Claude Code, Codex, Cursor, Copilot, Windsurf, OpenCode, Gemini CLI, and Antigravity. A minimal spec-driven-development harness: one command installs a 17-skill agentic SDLC (discovery, requirements, design, tasks, autonomous implementation). agent-skillsclaude-codecodexcursorgemini-cligithub-copilotkiroopencodesddspec-driven-development ✦ Claude review — Lower risk Kiro-inspired, well-scoped skills with plan-first defaults and per-task review; sampled skills are clean and structured. Markets 'long-running autonomous implementation', so use the review gates, but no risky surface and high heuristic score. INFO Autonomous implementation with built-in per-task review gates Heuristic signals Popular: 3,429 stars Licensed (MIT) Active (updated 13d ago) 256 forks Dependency manifests: 1 Has CI workflows Has SECURITY.md
agenticnotetaking/arscontextaNEW HooksPluginsResource ListsSkillsSubagents	3384	218	Shell	MIT	2026-02-24	✦ Moderate
Claude Code plugin that generates individualized knowledge systems from conversation. You describe how you think and work, have a conversation and get a complete second brain as markdown files you own. A plugin that generates a personalized 'second brain' knowledge system (folders, hooks, skills) from a conversation. claude-codeclaude-code-pluginknowledge-baseknowledge-managementmarkdownsecond-brain ✦ Claude review — Moderate Interesting generative approach producing markdown you own. The generated skills run Bash and use external research MCPs (Exa) plus WebSearch, and the setup generates and then activates its own hooks — so it writes and enables executable hooks on your machine. Review the generated hooks before activating. ~98 days stale. MED Generates and then activates its own hooks; skills run Bash and external research MCPs Heuristic signals Popular: 3,384 stars Licensed (MIT) Updated 98d ago 218 forks Contains install/shell scripts (7)
taishi-i/awesome-ChatGPT-repositoriesNEW PluginsResource ListsSkillsSlash Commands	3063	400	Python	CC0-1.0	2026-06-01	✦ Lower risk
A curated list of resources dedicated to open source GitHub repositories related to ChatGPT, OpenAI API, and Codex. Searchable via Claude Code skills. A curated, searchable list of 2,500+ ChatGPT/OpenAI/Codex open-source repos, queryable via a Claude skill. agentagent-skillsaiawesomeawesome-listchatgptcodexgpt-5llmopenai ✦ Claude review — Lower risk CC0-licensed index with a simple search skill over its own database; no risky execution. Clean and well-scoped. INFO Curated searchable index; CC0; simple search skill Heuristic signals Popular: 3,063 stars Licensed (CC0-1.0) Active (updated 1d ago) 400 forks Has CI workflows
Manavarya09/design-extractNEW MCP ServersPluginsSkillsSlash Commands	3022	272	JavaScript	MIT	2026-05-28	✦ Lower risk
Extract any website's complete design system with one command. DTCG tokens, semantic+primitive+composite, MCP server for Claude Code/Cursor/Windsurf, multi-platform emitters (iOS SwiftUI, Android Compose, Flutter, WordPress), Tailwind v4, Figma variables, shadcn/ui, CSS health audit, WCAG remediation, Chrome extension. MIT, Playwright, Node 20+. Extracts a website's full design system (DTCG tokens, Tailwind/Figma/shadcn output, WCAG audit) via an MCP and CLI. accessibilityagent-skillaichrome-extensionclaude-code-pluginclicsscursordesign-systemdesign-to-code ✦ Claude review — Lower risk Clean, well-scoped design tooling; the skill reads a URL and emits token files. Pulling design tokens from sites is far less fraught than cloning them, though respect site terms. One dev-setup script clones a PR branch (dev-only). INFO Extracts design tokens from arbitrary URLs; respect site terms Heuristic signals Popular: 3,022 stars Licensed (MIT) Active (updated 5d ago) 272 forks Dependency manifests: 4 Contains install/shell scripts (1) Has CI workflows Has SECURITY.md
davepoon/buildwithclaudeNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	3008	366	Python	MIT	2026-05-31	✦ Lower risk
A single hub to find Claude Skills, Agents, Commands, Hooks, Plugins, and Marketplace collections to extend Claude Code, Claude Desktop, Agent SDK and OpenClaw A plugin marketplace and discovery hub for Claude Code skills/agents/commands/hooks. claudeclaude-codeclaude-code-commandsclaude-skillscli-toolcommandsmcpmcp-servermcp-toolsopenclaw ✦ Claude review — Lower risk Marketplace plus sample 'agent-triforce' skills that delegate to named agents; sampled content is clean workflow guidance. 47 scripts across the marketplace; vet individual plugins you install from it. LOW Marketplace/discovery hub; vet individual plugins before installing Heuristic signals Popular: 3,008 stars Licensed (MIT) Active (updated 2d ago) 366 forks Dependency manifests: 6 Contains install/shell scripts (47) Has CI workflows
codeaashu/claude-codeNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	2787	3409	TypeScript	NOASSERTION	2026-04-22	✦ Elevated
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands. A repo distributing what it bills as the 'full leaked source' of Anthropic's Claude Code CLI (~1,900 files, 512k LOC). claudeclaude-aiclaude-codeclaude-code-leakedclaude-code-skillclaude-desktopclaude-leakclaude-skills ✦ Claude review — Elevated This is not a normal skills repo — it redistributes allegedly leaked proprietary Anthropic source code, with no clear license (NOASSERTION). Beyond the obvious legal/IP problems of using leaked proprietary code, you cannot trust that a third party's copy of a 512k-line codebase (with 36 install scripts) is unmodified. Avoid running it; use the official Claude Code instead. HIGH Redistributes allegedly leaked proprietary Anthropic source; legal/IP and trust risk MED No license; 36 install scripts in a huge third-party-hosted codebase you can't easily verify Heuristic signals Popular: 2,787 stars No clear license Updated 41d ago 3409 forks Dependency manifests: 3 Contains install/shell scripts (36)
giancarloerra/SocratiCodeNEW HooksMCP ServersPluginsSkillsSubagents	2778	371	TypeScript	AGPL-3.0	2026-05-27	✦ Lower risk
Enterprise-grade (40m+ LOC) codebase intelligence, zero-setup, local & private Plugin/Skill/Extension or MCP: hybrid semantic search, polyglot dependency graphs, symbol-level impact analysis & call-flow, interactive HTML viewer, cross-project & branch-aware search, DB/API/infra knowledge. 61% less tokens, 84% fewer calls, 37x faster. Cloud in beta. Local, private codebase-intelligence (hybrid semantic search, dependency graphs, impact analysis) via an MCP/plugin. aiai-assistantastclaudeclaude-codecode-graphcodebase-intelligencecontext-enginedockerembeddings ✦ Claude review — Lower risk AGPL-licensed, high heuristic score, sampled skills are read-and-index workflows that emphasize 'search before reading'. Local-first; clean. Standard indexing tool. INFO Local-first codebase indexing; clean Heuristic signals Popular: 2,778 stars Licensed (AGPL-3.0) Active (updated 6d ago) 371 forks Dependency manifests: 2 Has CI workflows Has SECURITY.md
notlikeDev/CCPluginsNEW Slash Commands	2709	157	Python	MIT	2025-10-07	✦ Moderate
Best Claude Code framework that actually save time. Built by a dev tired of typing "please act like a senior engineer" in every conversation. A slash-command framework of senior-engineer workflows (clean, commit, review, security-scan, etc.). automatedclaudeclaude-aiclaude-codeclicollectioncommandsextensionsplugins ✦ Claude review — Moderate The commands themselves are reasonable, but this repo is archived/unmaintained and its install.sh downloads command files at runtime from a different account's repo (brennercruvinel/CCPlugins raw URLs) — so what you install depends on a third-party source that could change. Prefer a maintained source. MED install.sh fetches command files from a different third-party repo at runtime LOW Repo is archived/unmaintained Heuristic signals Popular: 2,709 stars Licensed (MIT) Archived / unmaintained 157 forks README uses curl\|bash style install Contains install/shell scripts (4) Has CI workflows
PleasePrompto/notebooklm-mcpNEW MCP Servers	2669	375	TypeScript	MIT	2026-05-01	✦ Moderate
MCP server for NotebookLM - Let your AI agents (Claude Code, Codex) research documentation directly with grounded, citation-backed answers from Gemini. Persistent auth, library management, cross-client sharing. Zero hallucinations, just your knowledge base. An MCP server that drives a real Chrome (via Patchright stealth) to automate Google NotebookLM for grounded answers. ✦ Claude review — Moderate Useful for citation-backed research, but it works by stealth-automating a logged-in Google product with a 'persistent fingerprint' and multi-account support — that's anti-bot-evasion against Google's ToS, and it holds your Google session. Functional and popular, but go in aware of the ToS and credential exposure. MED Stealth browser automation of a logged-in Google product (ToS/anti-bot evasion); holds Google session Heuristic signals Popular: 2,669 stars Licensed (MIT) Updated 32d ago 375 forks Dependency manifests: 1
ZeframLou/call-meNEW HooksPluginsSkills	2598	251	TypeScript	None	2026-04-07	✦ Moderate
Minimal plugin that lets Claude Code call you on the phone. A plugin that lets Claude Code phone you (via Telnyx/Twilio + OpenAI speech + an ngrok tunnel) when it finishes, is blocked, or needs a decision. ✦ Claude review — Moderate Fun and well-scoped, but it requires telephony provider credentials and opens an ngrok webhook tunnel to your machine, and its Stop hook keeps Claude active to place a call. No clear license. The capability and the inbound tunnel are the things to understand before installing. MED Needs telephony + OpenAI credentials and opens an ngrok tunnel to your machine LOW Stop hook keeps the agent active to place calls; no clear license Heuristic signals Popular: 2,598 stars No clear license Updated 57d ago 251 forks Dependency manifests: 1
wshobson/commandsNEW General / Tooling	2495	285	—	MIT	2025-10-12	✦ Lower risk
A collection of production-ready slash commands for Claude Code A collection of 57 production-ready slash commands (workflows + tools) for Claude Code. aiai-agentsanthropicautomationclaudeclaude-codeclaude-commandsclaudecodeclaudecode-configorchestration ✦ Claude review — Lower risk Command/prompt definitions only; no risky execution surface in the scan. The author points users toward their newer plugin marketplace. Low risk. INFO Slash-command definitions; minimal surface Heuristic signals Popular: 2,495 stars Licensed (MIT) 285 forks
centminmod/my-claude-code-setupNEW HooksMCP ServersSkillsSlash CommandsSubagents	2384	227	Python	MIT	2026-06-01	✦ Lower risk
Shared starter template configuration and CLAUDE.md memory bank system for Claude Code A shared starter template and CLAUDE.md 'memory bank' system, with helper skills (image creation, session-metrics audit, docs consultant). claudeclaude-aiclaude-codeclaudecode-configclaudecode-hooksclaudecode-subagentssubagents ✦ Claude review — Lower risk Mostly templates and read-oriented skills; the image-creator uses OpenRouter via a Cloudflare AI Gateway BYOK setup (your keys), and the docs consultant fetches official Claude docs. Clean and conventional. INFO Templates plus BYOK image-gen and docs-fetch skills Heuristic signals Popular: 2,384 stars Licensed (MIT) Active (updated 1d ago) 227 forks Contains install/shell scripts (1)
jeremylongshore/claude-code-plugins-plus-skillsNEW HooksMCP ServersPluginsResource ListsSkillsSlash CommandsSubagents	2277	321	Python	MIT	2026-06-03	✦ Moderate
425 plugins, 2,810 skills, 200 agents for Claude Code. Open-source marketplace at tonsofskills.com with the ccpi CLI package manager. A very large marketplace (430+ plugins, 2,800+ skills, 200 agents) with its own 'ccpi' CLI package manager. agent-skillsaiai-agentsanthropicautomationclaude-codeclaude-code-pluginsdeveloper-toolsdevopsllm ✦ Claude review — Moderate Sampled skills (agency-os, amplify, audit) are clean and capable. The concern is scale and trust model: 531 install scripts, 479 manifests, and a dedicated package manager pulling a huge body of mixed-provenance skills. Install specific vetted items, not the whole catalog. MED Massive aggregated marketplace + custom package manager; 531 scripts / 479 manifests LOW Mixed-provenance skills; install selectively Heuristic signals Popular: 2,277 stars Licensed (MIT) Active (updated 0d ago) 321 forks Dependency manifests: 479 Contains install/shell scripts (531) Has CI workflows Has SECURITY.md
rohitg00/pro-workflowNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	2264	220	JavaScript	None	2026-06-03	✦ Lower risk
Claude Code learns from your corrections: self-correcting memory that compounds over 50+ sessions. Context engineering, parallel worktrees, agent teams, and 17 battle-tested skills. A self-correcting memory + workflow system: agent teams, parallel worktrees, batch orchestration, and 17 skills. agent-orchestrationai-agentsai-codingai-workflowclaudeclaude-codeclaude-code-pluginclaude-code-skillsclaude-skillscodex ✦ Claude review — Lower risk Sampled skills are sensible coordination/setup patterns (auto-detect project type, batch into worktrees with approval gates). It spawns parallel background agents in isolated worktrees by design. No license set, but no risky surface. LOW Spawns parallel background agents in worktrees; no clear license Heuristic signals Popular: 2,264 stars No clear license Active (updated 0d ago) 220 forks Dependency manifests: 1 Contains install/shell scripts (1) Has CI workflows
nizos/tdd-guardNEW HooksPluginsSkills	2168	166	TypeScript	MIT	2026-05-30	✦ Moderate
Automated TDD enforcement for Claude Code A plugin that enforces TDD by blocking Claude Code edits that skip tests or over-implement. agentic-codingautomationclaude-codecode-qualityhooksllm-toolstdd ✦ Claude review — Moderate Genuinely useful guardrail and it runs its own security CI. Note the mechanism: its hooks run `npx tdd-guard@latest` on PreToolUse, UserPromptSubmit, and SessionStart — so it executes the newest published npm version on essentially every action. That's auto-updating remote code execution baked into your session; pin a version if that concerns you. MED Hooks run `npx tdd-guard@latest` on every tool use/prompt/session (auto-updating remote execution) INFO Repo runs its own security CI Heuristic signals Popular: 2,168 stars Licensed (MIT) Active (updated 3d ago) 166 forks Dependency manifests: 17 Contains install/shell scripts (3) Has CI workflows
runkids/skillshareNEW HooksSkillsSlash Commands	2116	129	Go	MIT	2026-06-03	✦ Lower risk
📚 Sync skills across all AI CLI tools with one command and simplify team sharing. Supporting Codex, Claude Code, OpenClaw & more A Go CLI to sync skills across AI tools (Codex, Claude Code, OpenClaw) and simplify team sharing. agenthubaiclaude-codeclicodexcodex-skillscopilotcross-machine-synccursorgemini ✦ Claude review — Lower risk Sampled skills are read-only audits and E2E test runbooks (run in a devcontainer for isolation). 147 scripts but mostly its own test/release tooling; curl\|bash install. Sound, with a Go Report Card and security policy. LOW curl\|bash install; 147 scripts are largely its own test/release tooling Heuristic signals Popular: 2,116 stars Licensed (MIT) Active (updated 0d ago) 129 forks README uses curl\|bash style install Dependency manifests: 4 Contains install/shell scripts (147) Has CI workflows Has SECURITY.md
cyberagiinc/DevDocsNEW HooksMCP Servers	2083	192	TypeScript	Apache-2.0	2026-02-04	✦ Lower risk
Completely free, private, UI based Tech Documentation MCP server. Designed for coders and software developers in mind. Easily integrate into Cursor, Windsurf, Cline, Roo Code, Claude Desktop App A free, private, UI-based tech-documentation MCP server (a self-hosted alternative to doc-crawling services). clinecrawl4aicursordocumentationllmplaywrightpython3scrapersecuritytypescript ✦ Claude review — Lower risk Sampled code is ordinary React hooks. The README states it's not publicly maintained (an internal version is ahead) and it's ~118 days stale, but the surface is a self-hosted docs crawler/server. Low risk; just note maintenance status. LOW Not actively maintained publicly; ~118 days stale Heuristic signals Popular: 2,083 stars Licensed (Apache-2.0) Updated 118d ago 192 forks Dependency manifests: 3 Contains install/shell scripts (22)
greggh/claude-code.nvimNEW Hooks	2069	67	Lua	MIT	2026-02-04	✦ Lower risk
Seamless integration between Claude Code AI assistant and Neovim A Neovim plugin for integrating Claude Code into the editor. ai-assistantanthropicclaudeclaude-codeneovimnvimpluginterminal ✦ Claude review — Lower risk Clean Lua plugin; scripts are dev tooling (StyLua/luacheck git hooks, markdown fixers). MIT, has CI and SECURITY.md. ~118 days stale but low risk. INFO Editor integration; scripts are local dev tooling Heuristic signals Popular: 2,069 stars Licensed (MIT) Updated 118d ago 67 forks Contains install/shell scripts (7) Has CI workflows Has SECURITY.md
iannuttall/claude-agentsNEW Subagents	2058	274	—	MIT	2025-07-25	✦ Lower risk
Custom subagents to use with Claude Code. A small set of custom Claude Code subagents (refactorer, content-writer, security-auditor, etc.). ✦ Claude review — Lower risk Just markdown agent definitions copied into .claude/agents. Archived/unmaintained, but nothing executable or risky. INFO Markdown subagent definitions; archived but harmless Heuristic signals Popular: 2,058 stars Licensed (MIT) Archived / unmaintained 274 forks
yctimlin/mcp_excalidrawNEW MCP ServersSkills	2009	213	JavaScript	MIT	2026-05-16	✦ Lower risk
MCP server and Claude Code skill for Excalidraw — programmatic canvas toolkit to create, edit, and export diagrams via AI agents with real-time canvas sync. An MCP server and skill for programmatically driving a live Excalidraw canvas (create/edit/export diagrams). ✦ Claude review — Lower risk Clean diagramming tool; the skill talks to a local canvas server (default 127.0.0.1:3000). MIT, CI, Docker build. Low risk. INFO Drives a local Excalidraw canvas server; clean Heuristic signals Popular: 2,009 stars Licensed (MIT) Active (updated 17d ago) 213 forks Dependency manifests: 1 Has CI workflows
eugeniughelbur/obsidian-second-brainNEW HooksSkillsSlash Commands	1998	224	Python	MIT	2026-05-31	✦ Moderate
Cross-CLI skill for Obsidian: turn your vault into a living AI-first second brain across Claude Code, Codex, Gemini, and OpenCode. 43 commands - now with /obsidian-architect to document your codebase, key-less web research, Google Calendar, and self-rewriting notes. A cross-CLI Obsidian 'second brain' skill (43 commands) with self-rewriting notes, web research, and scheduled vault maintenance. ai-agentai-agentsai-automationai-researchai-toolsanthropicclaudeclaude-aiclaude-codeclaude-code-skill ✦ Claude review — Moderate Capable and the maintainer is careful: a SessionStart hook injects vault context (gated to the vault path), and an optional background agent that runs a headless Claude with --dangerously-skip-permissions ships INERT and trust-gated behind two env flags. Good defaults, but that skip-permissions background agent exists and writes unattended once enabled — understand it before turning it on. curl\|bash install. MED Opt-in background agent runs headless Claude with --dangerously-skip-permissions (ships disabled/trust-gated) LOW SessionStart hook injects vault context; curl\|bash install Heuristic signals Popular: 1,998 stars Licensed (MIT) Active (updated 2d ago) 224 forks README uses curl\|bash style install Dependency manifests: 1 Contains install/shell scripts (14) Has CI workflows Has SECURITY.md
wesammustafa/Claude-Code-Everything-You-Need-to-KnowNEW HooksMCP ServersSlash CommandsSubagents	1993	227	Python	MIT	2026-05-06	✦ Lower risk
The ultimate all-in-one guide to mastering Claude Code. From setup, prompt engineering, commands, hooks, workflows, automation, and integrations, to MCP servers, tools, and the BMAD method—packed with step-by-step tutorials, real-world examples, and expert strategies to make this the global go-to repo for Claude mastery. A comprehensive Claude Code guide with example hooks, MCP, workflows, and the BMAD method. ✦ Claude review — Lower risk Educational. Sampled hooks are reasonable and even defensive — pre_tool_use.py detects and can block dangerous `rm -rf` commands; notification hooks use optional TTS. uv-run single-file scripts. Low risk and good practices. INFO Educational; sample hooks include a defensive dangerous-command blocker Heuristic signals Popular: 1,993 stars Licensed (MIT) Active (updated 27d ago) 227 forks
glittercowboy/taches-cc-resourcesNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	1939	410	TypeScript	MIT	2026-04-01	✦ Lower risk
A collection of my favorite custom Claude Code resources to make life easier. A curated set of personal Claude Code resources: 27 commands and 9 skills (planning, meta-prompting, create-hooks/MCP). anthropicclaudeclaudecodecommandspromptsskillssubagents ✦ Claude review — Lower risk Sampled skills are meta-guidance for building skills/hooks/MCP servers — instructional, not auto-running. MIT, SECURITY.md. The 'assume everything is possible' philosophy is a mindset note, not a risk. Low. INFO Instructional skills for authoring skills/hooks/MCP Heuristic signals Popular: 1,939 stars Licensed (MIT) Updated 62d ago 410 forks Contains install/shell scripts (5) Has SECURITY.md
rohitg00/awesome-claude-code-toolkitNEW HooksMCP ServersPluginsResource ListsSkillsSlash CommandsSubagents	1931	610	JavaScript	Apache-2.0	2026-05-12	✦ Lower risk
The most comprehensive toolkit for Claude Code -- 135 agents, 35 curated skills, 42 commands, 176+ plugins, 20 hooks, 15 rules, 7 templates, 14 MCP configs, 26 companion apps, 52 ecosystem entries, and more. A broad toolkit/index: 135 agents, curated skills, commands, plugins, hooks, MCP configs, and ecosystem entries. claudeclaude-codeclaudecodeclaudecode-hookspluginsskills ✦ Claude review — Lower risk Sampled skills are clean reference content (WCAG, API design, an SEO orchestrator). Largely a curated aggregation with a curl\|bash install; vet linked third-party items and the optional external 'SkillKit' source. Apache-2.0. LOW Large curated aggregation; curl\|bash install; vet linked/external items Heuristic signals Popular: 1,931 stars Licensed (Apache-2.0) Active (updated 21d ago) 610 forks README uses curl\|bash style install Contains install/shell scripts (2) Has SECURITY.md
samber/cc-skills-golangNEW PluginsSkills	1924	125	Go	MIT	2026-05-29	✦ Lower risk
🧑‍🎨 A collection of Golang agentic skills that works A focused set of Go-specific agent skills (benchmarking, CLI design, code style, testing, security). agentagent-skillsaiantigravityclaudeclaude-codecodecodexcodingcopilot ✦ Claude review — Lower risk High-quality, human-reviewed Go guidance (the author explicitly distilled and reworked them, 'no AI slop'). Instructional content, well cross-referenced; no risky surface. Low risk. INFO Human-reviewed instructional Go skills Heuristic signals Popular: 1,924 stars Licensed (MIT) Active (updated 4d ago) 125 forks Contains install/shell scripts (1) Has CI workflows
jgravelle/jcodemunch-mcpNEW HooksMCP Servers	1884	294	Python	NOASSERTION	2026-05-30	✦ Moderate
The leading, most token-efficient MCP server for GitHub source code exploration via tree-sitter AST parsing A token-efficient MCP for GitHub/source-code exploration via tree-sitter AST parsing. claudeclaude-codeserenatokentoken-savingstokens ✦ Claude review — Moderate Capable code-navigation MCP. It leans on hooks that intercept the agent: a PreToolUse hook redirects Read of large code files to its own tools, a PostToolUse hook auto-reindexes after Edit/Write, and a worktree-event hook creates/removes git worktrees. One-click install runs a release wheel via uvx. Benign as written, but it inserts itself into your tool calls and filesystem; no clear license. MED Hooks intercept Read and auto-reindex on Edit/Write; manages git worktrees LOW One-click install runs a release wheel via uvx; no clear license Heuristic signals Popular: 1,884 stars No clear license Active (updated 3d ago) 294 forks Dependency manifests: 2 Contains install/shell scripts (8) Has CI workflows Has SECURITY.md
zubair-trabzada/ai-marketing-claudeNEW SkillsSubagents	1807	582	Python	MIT	2026-03-02	✦ Lower risk
AI Marketing Suite for Claude Code. 15 marketing skills with parallel subagents — audit any website, generate copy, email sequences, ad campaigns, content calendars, competitive intelligence, and client-ready PDF reports. A marketing suite of 15 skills that audit any website with parallel subagents and generate copy, ads, and client PDFs. ai-marketingclaudeclaude-codecompetitive-analysiscontent-strategycopywritingemail-marketingmarketingmarketing-automationseo ✦ Claude review — Lower risk Sampled skills fetch a target URL via WebFetch and run parallel analysis subagents — standard for a marketing auditor. curl\|bash install, ~93 days stale, but no risky surface beyond fetching public sites. Low risk. LOW Fetches arbitrary public URLs and runs parallel subagents; curl\|bash install Heuristic signals Popular: 1,807 stars Licensed (MIT) Updated 93d ago 582 forks README uses curl\|bash style install Dependency manifests: 1 Contains install/shell scripts (2)
tanbiralam/claude-codeNEW HooksMCP ServersPluginsSkillsSlash CommandsSubagents	1802	2578	TypeScript	None	2026-05-06	✦ Elevated
Claude Code is an agentic coding tool that lives in your terminal, understands your codebase, and helps you code faster by executing routine tasks, explaining complex code, and handling git workflows - all through natural language commands. All original source code is the property of Anthropic. Another repo redistributing the allegedly leaked source of Anthropic's Claude Code CLI (via an exposed npm .map file). claude-codeclaude-code-source-code ✦ Claude review — Elevated Same concern as the other 'leaked source' repo: it redistributes proprietary Anthropic code with no license and the README itself acknowledges the source belongs to Anthropic. Legal/IP exposure plus the impossibility of verifying a third-party copy (32 install scripts) make this one to avoid running; use official Claude Code. HIGH Redistributes allegedly leaked proprietary Anthropic source; legal/IP and trust risk MED No license; 32 install scripts; unverifiable third-party copy Heuristic signals Popular: 1,802 stars No clear license Active (updated 27d ago) 2578 forks README uses curl\|bash style install Dependency manifests: 7 Contains install/shell scripts (32)

✦ AI Plugin Radar · 100 repositories scanned · Claude review + heuristic signals · not a security audit — always review code before installing.

AI Plugin Radar

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Elevated

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Elevated

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Lower risk

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Moderate

Heuristic signals

✦ Claude review — Lower risk