# Claude Code Mods

> An unofficial community directory of Claude Code mods, plugins, skills, agents, hooks, commands and MCP servers. Each entry shows the date its source URL was last checked.

Claude Code Mods is not affiliated with or endorsed by Anthropic. "Source verified" means only that an entry's source URL responded on the date shown. It is not a security review, and no entry was scanned. Entries marked "Community listing" come from a third party and are not published by Anthropic: a maintainer read the entry, not the code. Mods are plugins whose behaviour lives in a hooks module. They are on by default in Claude Code v2.1.287 and later, and Anthropic ships some inside Claude Code. Catalog data was generated on 2026-10-10.

## Mods

- [agents-md](https://claudecodemods.com/extensions/agents-md/): Reads AGENTS.md as project instructions the way Claude Code reads CLAUDE.md, chosen by one option, instructionFiles, with four values. (Built in, by Anthropic)
- [diff](https://claudecodemods.com/extensions/diff/): Adds /diff: the session's uncommitted changes in a pane beside the transcript, file by file with their hunks, refreshed as Claude edits files and runs commands. (Built in, by Anthropic)
- [sec-default](https://claudecodemods.com/extensions/sec-default/): Keeps an organization's classic hooks, prompt content, managed settings and tool policy out of reach of user-installed plugins; adds no policy of its own. (Built in, by Anthropic)
- [telemetry](https://claudecodemods.com/extensions/telemetry/): Adds $.telemetry (log, mark) in the engine.create fold so built-in plugins can record first-party analytics rows, sent in batches; refuses installed plugins. (Built in, by Anthropic)
- [blast-radius](https://claudecodemods.com/extensions/blast-radius/): Anthropic's sample mod that holds a risky Bash command, such as rm -rf or a force push, and shows what it would change, with Proceed and Cancel. (Source only, by Anthropic)
- [Claude DevTools](https://claudecodemods.com/extensions/claude-devtools/): A debugger for Claude Code tool calls: breakpoints, pause/continue/reject/step, a timeline, an inspector, Error Lens failure diagnosis and sanitized exports. (Community listing, third-party text)
- [Mochi Crew](https://claudecodemods.com/extensions/mochi-crew/): A pixel-art cat in a side pane that reacts to your real session, with a kitten crew for subagents, XP and missions, a session meter and a live file tree. (Community listing, third-party text)
- [replay-theater](https://claudecodemods.com/extensions/replay-theater/): Anthropic's sample mod that adds /replay, a pane that steps through the file edits Claude made in the last turn, one diff at a time. (Source only, by Anthropic)
- [sweet-spot](https://claudecodemods.com/extensions/sweet-spot/): Hands a long session over before the context gets too big: to a fresh session that starts from a handoff, or by Claude Code's own /compact. (Community listing, third-party text)
- [TL;DR for Claude Code](https://claudecodemods.com/extensions/claude-tldr/): Every Claude Code turn as one card: files changed, tests before and after, and what to check. (Community listing, third-party text)
- [token-weather](https://claudecodemods.com/extensions/token-weather/): Anthropic's sample mod that draws a live forecast of the context window in the band above the prompt, with a chart of the last turns. (Source only, by Anthropic)

## Plugins

- [code-review](https://claudecodemods.com/extensions/code-review/): Automated pull request review with /code-review: four parallel agents, confidence scoring and an 80 point threshold. (Installable, by Anthropic)
- [commit-commands](https://claudecodemods.com/extensions/commit-commands/): Slash commands for git workflows: /commit, /commit-push-pr and /clean_gone, published in the anthropics/claude-plugins-official repository. (Installable, by Anthropic)
- [feature-dev](https://claudecodemods.com/extensions/feature-dev/): A guided 7-phase feature workflow behind /feature-dev, with code-explorer, code-architect and code-reviewer agents. (Installable, by Anthropic)
- [hookify](https://claudecodemods.com/extensions/hookify/): Create Claude Code hooks from markdown rule files instead of editing hooks.json, or derive rules from your conversation. (Installable, by Anthropic)
- [plugin-dev](https://claudecodemods.com/extensions/plugin-dev/): Toolkit for building Claude Code plugins: seven skills, three agents and an end-to-end /plugin-dev:create-plugin workflow. (Installable, by Anthropic)
- [pr-review-toolkit](https://claudecodemods.com/extensions/pr-review-toolkit/): Six review agents for pull requests (comments, tests, error handling, types, code quality, simplification) and a /review-pr command. (Installable, by Anthropic)
- [security-guidance](https://claudecodemods.com/extensions/security-guidance/): Three layers of security review for Claude-generated code: regex warnings on edits, an LLM diff review on Stop, and an agentic commit review. (Installable, by Anthropic)
- [session-report](https://claudecodemods.com/extensions/session-report/): Generates an explorable HTML report of Claude Code usage from local session transcripts: tokens, cache, subagents, skills, costly prompts. (Installable, by Anthropic)

## Skills

- [frontend-design](https://claudecodemods.com/extensions/frontend-design-skill/): A skill that steers Claude toward distinctive, intentional UI: aesthetic direction, typography, and avoiding templated defaults. (Installable, by Anthropic)
- [mcp-builder](https://claudecodemods.com/extensions/mcp-builder-skill/): A skill that guides building MCP servers in Python (FastMCP) or Node and TypeScript, from research through evaluations. (Installable, by Anthropic)
- [pdf](https://claudecodemods.com/extensions/pdf-skill/): A skill for reading, merging, splitting, watermarking, creating and OCRing PDFs, published as source-available reference. (Installable, by Anthropic)
- [skill-creator](https://claudecodemods.com/extensions/skill-creator-skill/): A skill for creating new skills, improving existing ones, running evals, and tuning a skill description for better triggering. (Installable, by Anthropic)
- [webapp-testing](https://claudecodemods.com/extensions/webapp-testing-skill/): A skill for testing local web apps with Playwright Python scripts, including a helper that starts and stops dev servers. (Installable, by Anthropic)
- [YYLO Skills](https://claudecodemods.com/extensions/yylo-skills/): Eight reusable agent skills for the YYLO agent workflow: Ledger task management and planning, project discovery, wiki and artifact records, and benchmark runs. (Community listing, third-party text)

## Agents

- [code-architect](https://claudecodemods.com/extensions/code-architect/): Read-only subagent from feature-dev that designs a feature blueprint from your existing code patterns and conventions. (Installable, by Anthropic)
- [code-explorer](https://claudecodemods.com/extensions/code-explorer/): Read-only subagent from feature-dev that traces how an existing feature works, from entry points through layers to data storage. (Installable, by Anthropic)
- [code-simplifier](https://claudecodemods.com/extensions/code-simplifier/): Subagent that simplifies recently modified code for clarity and consistency while preserving exact behavior. (Installable, by Anthropic)
- [silent-failure-hunter](https://claudecodemods.com/extensions/silent-failure-hunter/): Subagent from pr-review-toolkit that audits catch blocks, fallbacks and error handling for silent failures. (Installable, by Anthropic)
- [type-design-analyzer](https://claudecodemods.com/extensions/type-design-analyzer/): Subagent from pr-review-toolkit that reviews new or changed types for encapsulation, invariants, usefulness and enforcement. (Installable, by Anthropic)

## Hooks

- [explanatory-output-style SessionStart hook](https://claudecodemods.com/extensions/explanatory-style-session-hook/): A SessionStart hook that injects instructions asking Claude for short educational insights about implementation choices. (Installable, by Anthropic)
- [learning-output-style SessionStart hook](https://claudecodemods.com/extensions/learning-style-session-hook/): A SessionStart hook that switches Claude into a learning mode that asks you to write small, meaningful pieces of code. (Installable, by Anthropic)
- [ralph-loop Stop hook](https://claudecodemods.com/extensions/ralph-loop-stop-hook/): A Stop hook script that blocks Claude from exiting and feeds the same prompt back until a completion promise or an iteration cap is reached. (Installable, by Anthropic)

## Commands

- [/clean_gone](https://claudecodemods.com/extensions/clean-gone-command/): A slash command from commit-commands that deletes local git branches whose remote branch is gone, removing their worktrees first. (Installable, by Anthropic)
- [/new-sdk-app](https://claudecodemods.com/extensions/new-sdk-app-command/): A slash command from agent-sdk-dev that scaffolds a new Claude Agent SDK app in TypeScript or Python. (Installable, by Anthropic)
- [/review-pr](https://claudecodemods.com/extensions/review-pr-command/): A slash command from pr-review-toolkit that runs specialized review agents over your changes, optionally limited to chosen aspects. (Installable, by Anthropic)

## MCP servers

- [Context7 MCP](https://claudecodemods.com/extensions/context7-mcp/): Upstash's hosted MCP server that pulls current, version-specific library documentation into the model context. (Community listing, third-party text)
- [DC Hub](https://claudecodemods.com/extensions/dc-hub/): Remote MCP server with live data for data-center site selection, including DCPI, a daily score of where data centers can get grid power across 300+ markets. (Community listing, third-party text)
- [Fetch MCP server](https://claudecodemods.com/extensions/fetch-mcp-server/): Reference MCP server that fetches a URL and returns its content as markdown, with chunked reading via start_index. (Installable, by Model Context Protocol project)
- [Filesystem MCP server](https://claudecodemods.com/extensions/filesystem-mcp-server/): Reference MCP server for file operations with directory allowlists set by arguments or MCP Roots. (Installable, by Model Context Protocol project)
- [Git MCP server](https://claudecodemods.com/extensions/git-mcp-server/): Reference MCP server with tools to read, search and manipulate Git repositories, run with uvx mcp-server-git. (Installable, by Model Context Protocol project)
- [Memory MCP server](https://claudecodemods.com/extensions/memory-mcp-server/): Reference MCP server for persistent memory as a local knowledge graph of entities, relations and observations. (Installable, by Model Context Protocol project)
- [Playwright MCP](https://claudecodemods.com/extensions/playwright-mcp/): Microsoft's MCP server for browser automation through Playwright accessibility snapshots instead of screenshots. (Community listing, third-party text)
- [Sequential Thinking MCP server](https://claudecodemods.com/extensions/sequential-thinking-mcp-server/): Reference MCP server with one tool for step-by-step problem solving that supports revising thoughts and branching. (Installable, by Model Context Protocol project)

## Site docs

- [Mods](https://claudecodemods.com/mods/): Every Claude Code mod in the directory: the mods built into Claude Code, Anthropic's sample mods and community mods, with how to get each one.
- [Plugins](https://claudecodemods.com/plugins/): Claude Code plugins listed in the directory, with the /plugin install command for each one, what it adds to Claude Code and a link to its source.
- [Skills](https://claudecodemods.com/skills/): Claude Code skills in the directory: SKILL.md instructions that Claude loads when a task matches, with the plugin or source each one comes from.
- [Agents](https://claudecodemods.com/agents/): Claude Code subagents in the directory: agents with their own prompt, tools and context window, and the plugin you install to get each one.
- [MCP servers](https://claudecodemods.com/mcp-servers/): MCP servers that give Claude Code extra tools and data, with the claude mcp add or plugin command that sets up each one and a link to its source.
- [Learn](https://claudecodemods.com/learn/): What a Claude Mod is, how function hooks and the $ object work, and how mods differ from plugins, classic hooks, MCP servers and skills.
- [Getting started](https://claudecodemods.com/learn/getting-started/): Check that mods are on, ask Claude for a mod or build a starter mod that refuses force pushes, test it with claude plugin test and run it from a folder.
- [Migration](https://claudecodemods.com/learn/migration/): How a classic hook becomes a function hook: the classic.<Name> events, exit codes and JSON as returned objects, and a worked example.
- [Tutorials](https://claudecodemods.com/learn/tutorials/): The 9 official Claude Mods videos from the announcement issue, grouped Basic, Advanced and Case studies, played from GitHub's own attachment URLs.
- [Hooks](https://claudecodemods.com/hooks/): Every hook event Claude Code's function hooks name, by family and noun, with its source line and the mods, plugins and hooks that list it.
- [Security](https://claudecodemods.com/security/): What a mod can reach and how to review one before you install it, how organizations control mods, what is checked today and the proposed security tier model.
- [Publish](https://claudecodemods.com/publish/): List a Claude Code extension or mod by pull request: the steps, the rules a submission must follow, the entry format and what a listing means.
- [About](https://claudecodemods.com/about/): What this unofficial directory is, what a mod is, where the data comes from, what Source verified means and how to re-check the links.

## Proposed ideas (not real, no public implementation)

These were proposed in the marketplace spec and were last checked on 2026-09-20. None has a public repository, none can be installed, and none is in the directory above.

- [Security Guardrails and Secret Scrubber](https://claudecodemods.com/ideas/#claude-mod-guardrails): Would block destructive shell patterns and redact token-like strings from command arguments before a command or tool call runs.
- [Git Sentinel and Branch Hygiene](https://claudecodemods.com/ideas/#claude-mod-git-sentinel): Would keep Claude from editing or pushing to protected branches and ask for feature branches and structured commit summaries.
- [Smart Context Compressor and Token Pruner](https://claudecodemods.com/ideas/#claude-mod-context-compressor): Would deduplicate repeated tool calls and shorten large historical outputs when context is compacted or pruned.
- [Terminal HUD and Realtime Cost Sparkline](https://claudecodemods.com/ideas/#claude-mod-ui-telemetry): Would show a status line above the prompt with a token sparkline and the cumulative session cost.
- [Dynamic MCP Server-to-Hook Bridge](https://claudecodemods.com/ideas/#claude-mod-mcp-bridge): Would connect MCP servers to the hook system; the spec gives only its name, category and proposed events.
- [Ephemeral Docker Container Isolation](https://claudecodemods.com/ideas/#claude-mod-docker-sandbox): Would isolate command execution in an ephemeral Docker container; the spec gives only its name, category and proposed events.
- [LLM-as-a-Judge Pre-Commit Evaluator](https://claudecodemods.com/ideas/#claude-mod-agentic-eval): Would have an LLM judge evaluate work before a commit; the spec gives only its name, category and proposed events.
- [Terminal Audio and TTS Notifications](https://claudecodemods.com/ideas/#claude-mod-voice-alerts): Would play terminal audio and text to speech notifications; the spec gives only its name, category and proposed events.

## Optional

- [Browse all entries](https://claudecodemods.com/browse/): Search and filter the whole directory.
- [Full text of maintainer entries](https://claudecodemods.com/llms-full.txt): Guides and commands of maintainer and Anthropic entries, and an index line for each community entry.
- [Catalog JSON](https://claudecodemods.com/catalog.json): The whole catalog as one read-only JSON document. It is data, not a plugin marketplace.
- [Community marketplace](https://claudecodemods.com/marketplace.json): A Claude Code plugin marketplace of community plugins, each pinned to one commit. See https://claudecodemods.com/security/#marketplace for what a pin does and does not promise.
- [Atom feed](https://claudecodemods.com/feed.xml): The newest source checks.
- [Sitemap](https://claudecodemods.com/sitemap.xml): Every indexable page.
- [Source repository](https://github.com/joeVenner/claude-code-mods): Site source and community submissions by pull request.