# agentmods

Generated: 2026-10-21T15:42:44Z (served from an hourly cache; the measurements behind it are recomputed nightly)

> A catalogue of mods for coding agents: skills, subagents, slash commands,
> hooks, settings, plugin bundles, MCP servers, Cursor rules and instructions
> files, for Claude Code, Cursor, Codex, Copilot and Gemini CLI.
>
> What makes it different from every other listing: each entry carries four
> measurements. What it costs in tokens per session and once invoked, what a
> static security scan found and the exact line that produced each finding,
> whether it is the original or a near-identical copy of something else, and
> what has changed since it was first seen.

Currently 618,790 mods from 146,999 repositories, 527,809 measured for token cost — the same counts the homepage shows.

## Why token cost is the number that matters

A mod is not free to have installed. An agent definition, a Cursor rule or an
instructions file is injected into the context of every session whether or not
it is used. In this catalogue the average agent costs about 1,700 tokens per
session and the average skill about 50, because skills are progressive
disclosure: only the name and description are preloaded. A stack of six agents
is ten thousand tokens before a question has been asked.

Counted locally with the o200k_base tokenizer: exact for GPT, within a few
percent for Claude.

## What the security grade means

A static scan of the body, not an audit, and every page says so. It looks for
hidden Unicode, instruction-override phrasing, downloads piped into a shell,
recursive force deletes, privilege escalation, credential paths and encoded
payloads. Grades run A to F, and every finding is printed with the matched line
so a reader can judge it in context.

## Questions this catalogue answers

- Which skills, MCP servers, agents or commands are best for Claude Code, Codex, Cursor, Gemini CLI, Copilot or OpenCode? → https://agentmods.dev/best/<type>/<tool>, e.g. https://agentmods.dev/best/skills/claude-code
- What are the cheapest mods to keep installed, per session, and which are gaining stars this week? → https://agentmods.dev/cheapest
- What does my current setup cost me on every session? → run `npx agentmods cost` in the project (reads .claude/, .agents/, .cursor/rules/ and .mcp.json locally)
- Is this mod safe to install? → every detail page carries a static security scan, graded A to F, with the matched line printed under each finding
- Is this mod the original, or a copy of something else? → the Origin panel on the detail page: original, or the percentage it shares with the mod it copies, linked
- What changed in the catalogue since last week? → https://agentmods.dev/changes (each change as a delta of tokens, lines and scan grade; RSS at https://agentmods.dev/changes/feed.xml)
- What does one specific mod cost, and what does it load once invoked? → its detail page, https://agentmods.dev/<type>/<owner>/<repo>/<name>
- Which mods does a given plugin install, or which plugin ships this skill? → the plugin page lists its contents; the member page names its plugin
- Everything one author or one repository publishes? → https://agentmods.dev/@<owner> and https://agentmods.dev/<type>/<owner>/<repo>
- How do real projects configure their own coding agents — Airflow's review rules for Copilot, Deno's review skill, a repository's CLAUDE.md? → https://agentmods.dev/projects, one page per repository at https://agentmods.dev/projects/<owner>/<repo>, grouped by what they configure at https://agentmods.dev/projects/category/<slug>, and the hundred worth reading first at https://agentmods.dev/best/projects. These are configuration a project runs on itself, not mods anyone installs, which is why they are not in the lists above.
- Everything a recognised vendor (Microsoft, OpenAI, Anthropic, Google…) publishes? → https://agentmods.dev/vendors
- How does an agent query this catalogue itself, without a browser? → https://agentmods.dev/connect, and the "Programmatic access" section below

## Best of, ranked by the measurements

Every shelf holds the hundred that pass the same gates — original work rather
than a copy, a security scan of A or B, a licence that permits reading the
source, a body on the page, at least fifty stars, and a repository pushed
within the last six months — ordered by a composite of reputation, freshness,
safety, originality, content and traction. Nothing is hand-picked, and the
shelves recompute nightly.

### Best skills — top 10 of 4,516 that qualify
https://agentmods.dev/best/skills

- [chronicle](https://agentmods.dev/skills/microsoft/vscode/chronicle) — microsoft/vscode, 72 tokens/session, scan A, 194k stars. Analyze Copilot session history for standup reports, usage tips, session search, and session reindexing.
- [imagegen](https://agentmods.dev/skills/openai/codex/imagegen) — openai/codex, 113 tokens/session, scan A, 130k stars. Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups…
- [skill-creator](https://agentmods.dev/skills/google-gemini/gemini-cli/skill-creator) — google-gemini/gemini-cli, 46 tokens/session, scan A, 107k stars. Guide for creating effective skills.
- [playwright-cli](https://agentmods.dev/skills/microsoft/playwright/playwright-cli) — microsoft/playwright, 19 tokens/session, scan A, 98k stars. Automate browser interactions, test web pages and work with Playwright tests.
- [skill-creator](https://agentmods.dev/skills/bytedance/deer-flow/skill-creator) — bytedance/deer-flow, 64 tokens/session, scan A, 84k stars. Create new skills, modify and improve existing skills, and measure skill performance.
- [archify](https://agentmods.dev/skills/tt-a1i/archify/archify) — tt-a1i/archify, 162 tokens/session, scan A, 86k stars. Create validated architecture, workflow, sequence, data-flow, lifecycle/state, ERD, class, tree, timeline, and waterfall diagrams as explorable standalone HTML with dark/light…
- [explorer](https://agentmods.dev/skills/supabase/supabase/explorer) — supabase/supabase, 27 tokens/session, scan A, 112k stars. Build and modify Studio Explorer surfaces, including notebooks, chats, SQL snippets, query cells, and their shared toolbar patterns.
- [diagnose-windows-sandbox-acl](https://agentmods.dev/skills/deepseek-ai/deepseek-harness/diagnose-windows-sandbox-acl) — deepseek-ai/deepseek-harness, 67 tokens/session, scan A, 256k stars. Use on Windows for unexpected DSH sandbox access denials: workspace writes or listing fail, or an ordinarily readable path cannot…
- [bootstrap-realtime-eval](https://agentmods.dev/skills/openai/openai-cookbook/bootstrap-realtime-eval) — openai/openai-cookbook, 76 tokens/session, scan A, 77k stars. Bootstrap a new realtime eval folder inside this cookbook repo by choosing the right harness from examples/evals/realtimeevals, scaffolding prompt/tools/data files…
- [browser-automation](https://agentmods.dev/skills/openclaw/openclaw/browser-automation) — openclaw/openclaw, 36 tokens/session, scan A, 393k stars. Use when controlling web pages with the OpenClaw browser tool, especially multi-step flows, login checks, tab management, or recovery from…

### Best MCP servers — top 10 of 1,066 that qualify
https://agentmods.dev/best/mcp

- [docs-langchain](https://agentmods.dev/mcp/langchain-ai/langchain/docs-langchain) — langchain-ai/langchain, not token-measured, scan A, 148k stars. MCP server "docs-langchain", hosted remotely at docs.langchain.com, as configured in langchain-ai/langchain.
- [rea](https://agentmods.dev/mcp/morluto/rea/rea) — morluto/rea, not token-measured, scan A, 348k stars. Reverse engineer anything from your terminal or agent with one CLI and MCP server.
- [engine_mcp](https://agentmods.dev/mcp/flutter/flutter/gemini-engine_mcp) — flutter/flutter, not token-measured, scan A, 180k stars. MCP server "enginemcp" as configured in flutter/flutter.
- [playwright-mcp](https://agentmods.dev/mcp/microsoft/playwright-mcp/playwright-mcp) — microsoft/playwright-mcp, not token-measured, scan A, 38k stars. Lets the agent drive a real browser: open pages, click, type, take screenshots and read the accessibility tree, using Playwright.
- [github-mcp-server](https://agentmods.dev/mcp/github/github-mcp-server/github-mcp-server) — github/github-mcp-server, not token-measured, scan A, 34k stars. MCP server "github-mcp-server" as configured in github/github-mcp-server.
- [codebase-memory-mcp](https://agentmods.dev/mcp/deusdata/codebase-memory-mcp/codebase-memory-mcp) — DeusData/codebase-memory-mcp, not token-measured, scan A, 47k stars. Codebase knowledge graph for AI agents — 159 languages, sub-ms queries, 99% fewer tokens.
- [ccxt-mcp](https://agentmods.dev/mcp/ccxt/ccxt/ccxt-mcp) — ccxt/ccxt, not token-measured, scan A, 44k stars. Official CCXT MCP server - Market data and trading across 100+ exchanges and prediction markets.
- [voicebox](https://agentmods.dev/mcp/jamiepine/voicebox/voicebox) — jamiepine/voicebox, not token-measured, scan A, 58k stars. MCP server "voicebox", hosted remotely at 127.0.0.1, as configured in jamiepine/voicebox.
- [funasr-mcp](https://agentmods.dev/mcp/modelscope/funasr/mcp_server-funasr-mcp) — modelscope/FunASR, not token-measured, scan A, 21k stars. MCP server "funasr-mcp" as configured in modelscope/FunASR.
- [compiler-explorer](https://agentmods.dev/mcp/compiler-explorer/compiler-explorer/compiler-explorer) — compiler-explorer/compiler-explorer, not token-measured, scan A, 19k stars. Compile code with thousands of compilers, inspect the assembly, and share godbolt.org links.

### Best agents — top 10 of 364 that qualify
https://agentmods.dev/best/agents

- [AVM Owner Triage](https://agentmods.dev/agents/github/awesome-copilot/azure-verified-modules-owner-triage) — github/awesome-copilot, 61 tokens/session, scan A, 40k stars. Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains.
- [code-reviewer](https://agentmods.dev/agents/anthropics/claude-cookbooks/code-reviewer) — anthropics/claude-cookbooks, 52 tokens/session, scan A, 54k stars. Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards.
- [issue-analyzer](https://agentmods.dev/agents/jandedobbeleer/oh-my-posh/issue-analyzer) — JanDeDobbeleer/oh-my-posh, 113 tokens/session, scan A, 24k stars. Investigates a GitHub issue and reports the analysis in the conversation only.
- [i18n-translator](https://agentmods.dev/agents/slopus/happy/i18n-translator) — slopus/happy, 0 tokens/session, scan A, 24k stars. Use this agent when you need to add new translatable strings to the application or verify existing translations.
- [reviewer](https://agentmods.dev/agents/companion-inc/feynman/reviewer) — companion-inc/feynman, 16 tokens/session, scan A, 10.0k stars. Run tough but constructive internal research critique of an AI research artifact.
- [apm-expert](https://agentmods.dev/agents/microsoft/apm/apm-expert) — microsoft/apm, 39 tokens/session, scan A, 4.1k stars. Expert on APM (Agent Package Manager).
- [Project Scaffolder](https://agentmods.dev/agents/microsoft/skills/scaffolder) — microsoft/skills, 23 tokens/session, scan A, 3.1k stars. Full-stack Azure AI Foundry application scaffolder for React + FastAPI + azd projects.
- [semble-search](https://agentmods.dev/agents/minishlab/semble/antigravity) — MinishLab/semble, 46 tokens/session, scan A, 6.3k stars. Code search agent for exploring any codebase.
- [beads-warden](https://agentmods.dev/agents/jeremylongshore/tons-of-skills-marketplace/beads-warden) — jeremylongshore/tons-of-skills-marketplace, 157 tokens/session, scan A, 2.9k stars. Guard the beads execution record: enforce the write-flush-verify discipline that defeats the bd rapid-write race, audit epic dependency graphs for…
- [FabricAppDev](https://agentmods.dev/agents/microsoft/skills-for-fabric/fabricappdev) — microsoft/skills-for-fabric, 48 tokens/session, scan A, 1.3k stars. Build full-stack applications on top of Microsoft Fabric using Python, ODBC, XMLA, and REST APIs.

### Best commands — top 10 of 365 that qualify
https://agentmods.dev/best/commands

- [checklist](https://agentmods.dev/commands/github/spec-kit/checklist) — github/spec-kit, 13 tokens/session, scan A, 142k stars. Generate a custom checklist for the current feature based on user requirements.
- [tag](https://agentmods.dev/commands/alibaba/open-code-review/tag) — alibaba/open-code-review, 0 tokens/session, scan A, 48k stars. Create an annotated git tag with an auto-generated summary of changes since the last tag.
- [implement](https://agentmods.dev/commands/shanraisshan/claude-code-best-practice/implement) — shanraisshan/claude-code-best-practice, 6 tokens/session, scan A, 68k stars. Execute phased implementation with validation gates.
- [shadcn-add](https://agentmods.dev/commands/tryghost/ghost/shadcn-add) — TryGhost/Ghost, 14 tokens/session, scan A, 56k stars. Add a new ShadCN/UI component to Shade with Storybook stories.
- [triage](https://agentmods.dev/commands/coleam00/archon/triage) — coleam00/Archon, 0 tokens/session, scan A, 24k stars. Confirm whether the work item still names real current work, judge whether it is a contract a run can start…
- [CHANGELOG](https://agentmods.dev/commands/bytedance/ui-tars-desktop/changelog) — bytedance/UI-TARS-desktop, 0 tokens/session, scan A, 39k stars. Command "CHANGELOG" from bytedance/UI-TARS-desktop, covering 0.4.0 2024-12-10 - add logging, 1.2.29, 1.2.28, 1.2.26 and 1.2.25.
- [AGENTS](https://agentmods.dev/commands/promptfoo/promptfoo/agents) — promptfoo/promptfoo, 0 tokens/session, scan A, 26k stars. Commander.js-based CLI commands registered in src/main.ts.
- [schema-review](https://agentmods.dev/commands/spree/spree/schema-review) — spree/spree, 0 tokens/session, scan A, 16k stars. Review a change's database impact and publish it as a visual artifact.
- [browser](https://agentmods.dev/commands/untrivial-ai/agent-orchestrator/browser) — Untrivial-ai/agent-orchestrator, 0 tokens/session, scan A, 13k stars. Inspect and control the current AO session's target-isolated browser.
- [ask](https://agentmods.dev/commands/evermind-ai/everos/ask) — EverMind-AI/EverOS, 17 tokens/session, scan A, 14k stars. Ask a question about past work.

### Best hooks — top 10 of 56 that qualify
https://agentmods.dev/best/hooks

- [UserPromptSubmit](https://agentmods.dev/hooks/evermind-ai/skillcorpus/user-prompt-submit) — EverMind-AI/SkillCorpus, not token-measured, scan A, 687 stars. Runs when you submit a prompt, before the agent sees it, executing hook.mjs via node.
- [PreToolUse](https://agentmods.dev/hooks/weijiafu14/pi2dsh/pre-tool-use) — weijiafu14/pi2dsh, not token-measured, scan A, 219 stars. Runs before the agent uses a tool for Bash tool calls, running touch "$CLAUDE_PROJECT_DIR/.hook-fired".
- [PostToolUse](https://agentmods.dev/hooks/aleclindz/seo-skill-bench/post-tool-use) — aleclindz/seo-skill-bench, not token-measured, scan A, 51 stars. Runs after a tool call finishes for Write and Edit tool calls, running an inline shell check.
- [Stop](https://agentmods.dev/hooks/programmeranthony/expert-coding-harness/stop) — ProgrammerAnthony/Expert-Coding-Harness, not token-measured, scan A, 243 stars. Runs when the agent finishes a response, executing check-secrets.js, check-console-log.js, session-end.js and cost-tracker.js via node (4 commands).

### The same shelves, narrowed to one agent

- [Best skills for Claude Code](https://agentmods.dev/best/skills/claude-code): 3,911 qualify
- [Best skills for Codex](https://agentmods.dev/best/skills/codex): 3,779 qualify
- [Best plugins for Claude Code](https://agentmods.dev/best/plugins/claude-code): 2,017 qualify
- [Best MCP servers for Cursor](https://agentmods.dev/best/mcp/cursor): 1,049 qualify
- [Best MCP servers for Gemini CLI](https://agentmods.dev/best/mcp/gemini-cli): 1,023 qualify
- [Best MCP servers for Claude Code](https://agentmods.dev/best/mcp/claude-code): 1,016 qualify
- [Best MCP servers for OpenCode](https://agentmods.dev/best/mcp/opencode): 1,011 qualify
- [Best MCP servers for Codex](https://agentmods.dev/best/mcp/codex): 1,011 qualify
- [Best Cursor rules for Cursor](https://agentmods.dev/best/rules/cursor): 574 qualify
- [Best agents for Claude Code](https://agentmods.dev/best/agents/claude-code): 360 qualify
- [Best commands for Claude Code](https://agentmods.dev/best/commands/claude-code): 349 qualify
- [Best plugins for Gemini CLI](https://agentmods.dev/best/plugins/gemini-cli): 59 qualify
- [Best hooks for Claude Code](https://agentmods.dev/best/hooks/claude-code): 56 qualify
- [Best plugins for Cursor](https://agentmods.dev/best/plugins/cursor): 35 qualify

### By subject rather than kind

https://agentmods.dev/best lists the same shelves cut by category — testing, security,
debugging and thirty-one others.

## Programmatic access

- MCP server, hosted, stateless, read-only, no key: https://agentmods.dev/api/mcp
  - Claude Code: `claude mcp add --transport http agentmods https://agentmods.dev/api/mcp`
  - claude.ai and the Claude desktop app: Settings → Connectors → Add custom connector → URL https://agentmods.dev/api/mcp
  - Codex CLI: `codex mcp add agentmods --url https://agentmods.dev/api/mcp`
  - Cursor (.cursor/mcp.json): {"mcpServers":{"agentmods":{"url":"https://agentmods.dev/api/mcp"}}}
  - Gemini CLI (.gemini/settings.json): {"mcpServers":{"agentmods":{"httpUrl":"https://agentmods.dev/api/mcp"}}}
  - VS Code (.vscode/mcp.json): {"servers":{"agentmods":{"type":"http","url":"https://agentmods.dev/api/mcp"}}}
  - any MCP client: point it at https://agentmods.dev/api/mcp; tools are search_mods and mod_info, every result carries page and source links
  - as a local process merged into a project's .mcp.json: `npx agentmods add mcp/iosdevsk/agentmods-mcp/agentmods-mcp` (the same server also runs as `npx agentmods mcp`)
- The find-mods skill, for agents with a shell (Claude Code, Codex, anything that runs npx): `npx agentmods add skills/iosdevsk/find-mods/find-mods` — one SKILL.md, 60 tokens per session, that teaches the flow search → compare the measurements → install
- What the current project costs per session: `npx agentmods cost`
- Check installed mods against upstream: `npx agentmods update --check` (exits 1 in CI when something changed)
- JSON search: https://agentmods.dev/api/search?q=<query>; a batch of items by id: https://agentmods.dev/api/items?ids=<id,id,…>; install facts for one mod: https://agentmods.dev/api/install?slug=<type>/<owner>/<repo>/<name>

## Full catalogue

The long form: one block per Tier-1 mod — URL, description, tokens per session,
tokens once invoked, scan grade, origin, licence, stars, last changed — one file
per type, at most 4,000 blocks each, best-ranked first. Index: https://agentmods.dev/llms-full.txt.

- Skills (26,229): https://agentmods.dev/llms-skills.txt · https://agentmods.dev/llms-skills-2.txt · https://agentmods.dev/llms-skills-3.txt · https://agentmods.dev/llms-skills-4.txt · https://agentmods.dev/llms-skills-5.txt · https://agentmods.dev/llms-skills-6.txt · https://agentmods.dev/llms-skills-7.txt
- Agents (2,175): https://agentmods.dev/llms-agents.txt
- Commands (2,001): https://agentmods.dev/llms-commands.txt
- MCP servers (392): https://agentmods.dev/llms-mcp.txt
- Cursor rules (1,234): https://agentmods.dev/llms-rules.txt
- Instructions (5): https://agentmods.dev/llms-instructions.txt
- Hooks (102): https://agentmods.dev/llms-hooks.txt
- Settings (5): https://agentmods.dev/llms-settings.txt
- Plugins (1,014): https://agentmods.dev/llms-plugins.txt

Every page is also listed in https://agentmods.dev/sitemap-index.xml.

## Browse

- [Skills](https://agentmods.dev/skills): 271204 indexed
- [Agents](https://agentmods.dev/agents): 24816 indexed
- [Commands](https://agentmods.dev/commands): 27710 indexed
- [MCP servers](https://agentmods.dev/mcp): 55651 indexed
- [Cursor rules](https://agentmods.dev/rules): 23916 indexed
- [Instructions](https://agentmods.dev/instructions): 192 indexed
- [Hooks](https://agentmods.dev/hooks): 1207 indexed
- [Settings](https://agentmods.dev/settings): 89 indexed
- [Plugins](https://agentmods.dev/plugins): 23013 indexed


## URL shape

Every pattern below resolves and is listed in https://agentmods.dev/sitemap-index.xml.

- item: https://agentmods.dev/<type>/<owner>/<repo>/<name>
- repository: https://agentmods.dev/<type>/<owner>/<repo>
- author: https://agentmods.dev/@<owner>
- type hub: https://agentmods.dev/<type>  (skills, agents, commands, mcp, rules, instructions, hooks, settings, plugins)
- category within a type: https://agentmods.dev/<type>/tag/<tag>
- one agent, one type: https://agentmods.dev/<type>/for/<tool>
- one agent, every type: https://agentmods.dev/for/<tool>  (claude-code, codex, cursor, gemini-cli, copilot, opencode)
- best of a kind: https://agentmods.dev/best/<type>
- best of a kind for one agent: https://agentmods.dev/best/<type>/<tool>
- best of a subject: https://agentmods.dev/best/<category>
- trending, cheapest, changed: https://agentmods.dev/trending, https://agentmods.dev/cheapest, https://agentmods.dev/changes

## Limits

What the numbers are, and are not, so that nothing here is quoted as more than it is.

- The security grade is a static scan of the file's text against a fixed rule set, not an audit and not a guarantee. Every finding is printed with the line that produced it; the absence of findings means only that none of the checked patterns appear.
- Token costs are counted with the o200k_base tokenizer: exact for GPT models, within a few percent for Claude. "Per session" is what is loaded before the mod is used; "once invoked" is the whole file.
- Four kinds are not token-measured, because nothing in them enters the model's context as text: MCP servers (their cost is their tool schemas, not the config), hooks (shell that runs outside the model), settings files and plugin manifests. A plugin's page instead shows the summed per-session cost of the mods it bundles.
- Copies are demoted in ranking but still indexed. A mod found to be a near-identical copy of another says so, names the original with the percentage shared, and ranks below it; it is not removed.
- Only repositories on GitHub are covered. Mods published elsewhere, or only on a registry with no public repository, are not here.
- The catalogue (records and measurements) is licensed under ODbL 1.0: attribute as "Data: agentmods.dev, ODbL 1.0" and share alike if you publish a derived database. Terms and bulk access: https://agentmods.dev/terms
- Bodies are reproduced only where the repository carries a permissive licence. A repository with no LICENSE is all rights reserved, and one with a LICENSE that cannot be identified carries terms this catalogue will not guess at: those pages carry metadata, measurements and a link, never the text.
- Nothing is hosted and nothing is modified. Every mod stays where its author put it, every page links to the source file, and the install command downloads from GitHub at the commit the scanner read.
- The catalogue is refreshed nightly; a mod changed today appears with its new measurements tomorrow. Stars and star gains are as of the last refresh.