接入你的智能体
从发布者处安装。不是 PlugMCP 端点。
远程 URL 还是本机命令? 两种安装方式的区别
A free, community-maintained directory of plugins in Anthropic's official Claude Plugins catalog — installable bundles of skills, MCP servers, slash commands, sub-agents, and hooks that extend Claude Code and Cowork with one-command installation.
[!NOTE] This is a free, independent, community-maintained list. Not affiliated with, endorsed by, or sponsored by Anthropic PBC. "Claude," "Claude Code," and "Cowork" are trademarks of Anthropic PBC. Each plugin is the property of its respective owner. This list is published under the CC0 license shown in the badge above — public domain, free to copy, fork, redistribute.
Last updated: July 23, 2026 | Total plugins: 348 | Surfaces: Claude Code (286) · Cowork (125) · both (63) | Categories: 28
Plugins originated in Claude Code (October 2025) as packaged, versioned, shareable directories that bundle skills, MCP server references, slash commands, sub-agents, hooks, and LSP servers into a single installable unit. As of v2.1.129 (May 6, 2026), Anthropic runs the Official Claude Plugins Directory (claude-plugins-official) — a curated, vetted catalog auto-installed on Claude Code startup. Plugins also extend Cowork, Anthropic's agentic workspace, with role-class workflow bundles (Productivity, Design, Marketing, etc.) used by knowledge workers inside isolated VM environments.
For more information, see the Plugins overview, plugin marketplaces guide, submission form, and the official Anthropic marketplace repo (renamed from claude-plugins-public; the old URL still redirects).
Plugins marked with A are built and verified by Anthropic. Each entry ends with a surface tag: (Claude Code), (Cowork), or (Claude Code, Cowork).
This list is maintained weekly. To contribute, see CONTRIBUTING.md.
[!NOTE]
Two surfaces, one list — why this is bigger than the claude.ai plugin browser
Plugins run on two surfaces, and this list catalogs both — the surface tag on every entry tells you where each one is available:
- Claude Code (the developer CLI) — its marketplace (
claude-plugins-official) is the full catalog, auto-installed on startup and dominated by developer and infrastructure plugins: language servers, cloud (AWS/GCP/Azure/Oracle), databases, CI/CD, security, and code tooling. ~286 of the entries below run here.- Cowork (Anthropic's agentic workspace inside claude.ai) — its plugin library surfaces a curated business/role subset: CRM, finance, marketing, design, data dashboards, the Anthropic role bundles, and — new this week — full Legal (12 role-specialized plugins) and Healthcare / life-sciences verticals. 125 entries run here; 63 run on both.
So the plugin browser in the claude.ai desktop app shows far fewer plugins than this list — it's the Cowork storefront, not the whole catalog. Developer and infrastructure plugins don't appear there because they target a coding environment, not a Cowork VM. If a plugin you expected is missing from the desktop app, it's almost certainly a Claude Code–only entry.
Scope note: counts here track the official
claude-plugins-officialmarketplace plus the Cowork directory. A handful of plugins ship from partner marketplaces (e.g.techwolf-ai/ai-first-toolkit) and appear in the public Claude Plugins catalog but not the official marketplace manifest; those are added individually as they're found.
[!IMPORTANT]
Plugin vs Connector vs MCP vs Custom Plugin
These four concepts overlap and confuse people. Here's the clean mental model:
- MCP (Model Context Protocol) is the open standard. Servers expose tools, prompts, and resources; clients call them. Spec at modelcontextprotocol.io.
- MCP server is any running server implementing MCP. Remote MCP servers (internet-hosted, OAuth) work on every Claude surface; local MCP servers (stdio) work only in Claude Desktop and Claude Code.
- Connector is an MCP server listed in Anthropic's Connectors Directory — reviewed by Anthropic, named-card UI, cross-surface (Claude.ai, Desktop, Mobile, Cowork, Code). Tracked in awesome-claude-connectors.
- Custom Connector is an MCP server you add yourself via Settings → "Add custom connector" with a URL. Same runtime as directory connectors; no Anthropic review; appears as "Custom".
- Plugin is a packaged bundle that can contain any combination of skills, MCP server references, slash commands, sub-agents, hooks, and LSP servers. Versioned, semver, one-command install via
/plugin install. Works on Claude Code and Cowork. This list covers plugins.- Plugin Marketplace is a Git repo or URL hosting
.claude-plugin/marketplace.jsonthat catalogs plugins. The official Anthropic marketplace (claude-plugins-official) is auto-installed on Claude Code startup.- Custom Plugin is a plugin you build and install from a local directory, your own Git repo, or a private marketplace. Same components as directory plugins; no Anthropic review. Install via
/plugin install,--plugin-dir, or--plugin-url.Mental model: Connector = "Claude can call your API" (tool surface). Plugin = "Claude knows how to use your product" (workflow + tools + skills). MCP = the protocol both speak. Most partners ship both — a remote MCP server (connector) and a plugin that wraps it with skills and slash commands.
[!TIP]
Plugin of the Week — July 23, 2026
Litigation Legal · Legal
The story this week is a vertical, not a single plugin: Cowork gained an entire Legal practice suite — twelve role-specialized, Anthropic-built plugins covering Litigation, Corporate, IP, Employment, Privacy, Regulatory, Commercial, Product, and AI Governance work, plus Law Student, Legal Clinic (built within ABA Formal Op. 512), and a security-gated Legal Builder Hub. Litigation Legal headlines it: it manages the matter portfolio — deadlines, holds, demands, outside counsel — and does the production work too, from patent and civil claim charts to chronologies, deposition prep, privilege logs, and brief drafts, adapting to in-house, firm, or solo practice. Landing alongside it is a deep Healthcare and life-sciences cohort — Claude for Healthcare (prior auth, ICD-10-CM coding, FHIR) and a research bench of Open Targets, Owkin, Synapse, ToolUniverse, scvi-tools, Medidata, Cortellis, BioRender, ChEMBL, and more. This is the largest single reconcile in the list's history: +54 plugins (279 → 333), almost all of it Cowork's business and professional verticals rather than the developer marketplace, which added a smaller wave of its own (New Relic, MLflow, GrowthBook, GitKraken, DeepEval, CrowdSec, Browser Use, and Anthropic's own Claude Security and Receipts). No plugins were removed: three that left the official manifest (Adspirer, SearchFit, Product Tracking) remain live in Cowork and were re-tagged Cowork-only rather than dropped. Listed under Legal below.
Inside Claude Code, type /plugin to open the Plugin Manager. The official Anthropic marketplace (claude-plugins-official) is auto-installed on startup, so most plugins below are one command away:
/plugin install <plugin-name>@claude-plugins-official
For plugins from other marketplaces, add the marketplace first:
/plugin marketplace add owner/repo
/plugin install <plugin-name>
To load a plugin from a local directory or remote .zip (for one-off testing or private builds):
claude --plugin-dir /path/to/plugin
claude --plugin-url https://example.com/plugin.zip
In Cowork, plugins appear in the workspace's plugin manager — connect the marketplace once per workspace and enable plugins per project.
A - Development kit for working with the Claude Agent SDK. Use case: Scaffolding agents, validating SDK usage, testing agent workflows locally. (Claude Code)A - Search across all of your company's tools in one place — email, chat, documents, and wikis without switching between apps. Use case: Cross-tool knowledge retrieval, finding decisions across Slack/email/Drive, answering "where did we discuss X" questions. (Cowork)A - Optimize business operations including vendor management, process documentation, change management, capacity planning, and compliance tracking. Use case: Vendor onboarding, SOP authoring, ops dashboards, audit prep. (Cowork)A - Pre-built small business workflows including payroll planning, month-end close, weekly briefs, and growth campaigns using QuickBooks, PayPal, HubSpot, and related tools. Use case: Solo founder ops, bookkeeping triage, weekly business reviews, growth-campaign coordination. (Cowork)A - AI code review with specialized agents and confidence-based filtering for pull requests. Use case: PR review automation, multi-agent code analysis, confidence-weighted feedback. (Claude Code)A - Code clarity agent that simplifies and refines recently modified code while preserving functionality and consistency. Use case: Post-edit cleanup, complexity reduction, dead-code elimination after refactors. (Claude Code)A - PR review agents for comments, tests, errors, types, quality, and simplification. Use case: Multi-axis PR review, test-coverage check, type-correctness validation in pull requests. (Claude Code)A - Generate an explorable HTML report of Claude Code session usage including tokens, cache efficiency, subagents, skills, and tool calls. Use case: Session retrospection, cache-hit analysis, subagent usage audit. (Claude Code)A - Official Slack MCP server with skills and slash commands for searching messages, sending communications, managing canvases, and surfacing channel insights. Use case: Slack-message drafting, channel-thread summarization, async standups, in-Slack analysis. (Claude Code, Cowork)A - Prospect, craft outreach, build deal strategy, prep for calls, manage pipeline, and write personalized messaging. Use case: Pipeline management, prospect research, personalized cold outreach. (Cowork)A - Triage tickets, draft responses, escalate issues, build knowledge base, research customer context, and turn resolved issues into self-service content. Use case: Support-ticket triage, response drafting, knowledge-base content generation. (Cowork)A - Write SQL, explore datasets, generate insights, build visualizations and dashboards, and turn raw data into clear stories. Use case: Ad-hoc SQL, dashboard authoring, stakeholder-facing data storytelling. (Claude Code, Cowork)sup CLI for shell, scripting, and CI/CD workflows. Use case: Superset/Preset automation, BI-asset CI/CD, scripted dashboard management. (Claude Code)A - Accelerate design workflows including critique, design system management, UX writing, accessibility audits, research synthesis, and dev handoff. Use case: Design critique, design-system component authoring, accessibility audit, design-to-dev handoff specs. (Cowork)A - Craft production-grade frontends with distinctive design that avoids generic AI aesthetics. Use case: Landing-page design, app-shell composition, distinctive component design, anti-shadcn-default polish. (Claude Code)A - Interactive HTML playgrounds with visual controls, live preview, and prompt output including design, data, concept map, and critique templates. Use case: Live design iteration, prompt-output visualization, concept-map authoring, in-chat critique surfaces. (Claude Code)A - Analyze codebases and recommend tailored Claude Code automations including hooks, skills, MCP servers, and subagents. Use case: First-time Claude Code setup on an unfamiliar repo, automation discovery, configuration audit. (Claude Code)A - Maintain CLAUDE.md by auditing quality, capturing learnings, and keeping project memory current. Use case: CLAUDE.md hygiene, learning capture, project-memory refresh after major changes. (Claude Code)A - Modernize legacy codebases such as COBOL, legacy Java/C++, and monoliths with a structured assess, map, extract, and transform workflow. Use case: Legacy-code assessment, business-rule extraction, staged modernization. (Claude Code)A - Commands for Git commit workflows including commit, push, and PR creation. Use case: Commit-message generation, PR creation, conventional-commit enforcement. (Claude Code)A - Onboard a Code-with-Claude Makers Cardputer with one command that clones the repo, flashes firmware, and installs the Claude Buddy app. Use case: Cardputer setup, UIFlow firmware flashing, maker-project onboarding. (Claude Code)A - Feature development workflow with agents for exploration, design, and review. Use case: New-feature scoping, multi-phase development workflow, design-then-review cadence. (Claude Code)A - Create custom hooks via markdown to prevent unwanted behaviors from conversation patterns or explicit instructions. Use case: Behavioral guardrail authoring, custom Stop hooks, pattern-based hook generation. (Claude Code)A - Interactive AI loops for iterative development using the Ralph Wiggum technique where Claude works on tasks repeatedly until completion. Use case: Stubborn-task iteration, /loop-driven development, autonomous task completion. (Claude Code)A - Solve competition math (IMO, Putnam, USAMO) with adversarial verification that attacks proofs to catch errors self-verification misses. Use case: Olympiad proof solving, adversarial proof verification, calibrated abstention. (Claude Code)A - Streamline engineering workflows including standups, code review, architecture decisions, incident response, and technical documentation. Use case: Engineering standups, architecture-decision documentation, incident post-mortems. (Cowork)A - Write feature specs, plan roadmaps, synthesize user research, keep stakeholders updated, and stay ahead of competitive landscape. Use case: Spec authoring, roadmap planning, research synthesis, competitive monitoring. (Cowork)A - Streamline finance and accounting workflows from journal entries and reconciliation to financial statements, variance analysis, audit prep, and month-end close. Use case: Month-end close, variance analysis, audit prep, journal-entry triage. (Cowork)A - Connect to preclinical research tools and databases including literature search, genomics analysis, and target prioritization. Use case: Early-stage life sciences R&D, target discovery, preclinical literature synthesis. (Cowork)A - Prior authorization, ICD-10-CM medical coding, clinical-trial protocols, and FHIR development, with CMS Coverage, ICD-10, NPI Registry, Clinical Trials, and PubMed connected. Use case: Prior-auth drafting, ICD-10-CM coding, clinical-trial protocol and FHIR work. (Cowork)A - Streamline people operations including recruiting, onboarding, performance reviews, compensation analysis, and policy guidance. Use case: Recruiting workflows, performance-review drafting, compensation benchmarking, HR-policy lookups. (Cowork)A - C# language server for code intelligence. Use case: C#/.NET autocomplete, type checking, refactoring support inside Claude Code. (Claude Code)A - C/C++ language server (clangd) for code intelligence. Use case: C/C++ autocomplete, diagnostics, cross-reference navigation. (Claude Code)A - Go language server for code intelligence and refactoring. Use case: Go autocomplete, refactor-rename support, package navigation. (Claude Code)A - Java language server (Eclipse JDT.LS) for code intelligence. Use case: Java autocomplete, type checking, JDT-powered refactoring. (Claude Code)A - Kotlin language server for code intelligence. Use case: Kotlin autocomplete, Android development, multiplatform Kotlin work. (Claude Code)A - Lua language server for code intelligence. Use case: Lua scripting, Neovim configuration, game-engine Lua editing. (Claude Code)A - PHP language server (Intelephense) for code intelligence. Use case: PHP autocomplete, Laravel/Symfony development, type-inference assistance. (Claude Code)A - Python language server (Pyright) for type checking and code intelligence. Use case: Python type checking, autocomplete, strict-mode type inference. (Claude Code)A - Ruby language server for code intelligence and analysis. Use case: Ruby/Rails autocomplete, diagnostic feedback, refactor support. (Claude Code)A - Rust language server for code intelligence and analysis. Use case: Rust autocomplete, type inference, refactor-rename support. (Claude Code)A - Swift language server (SourceKit-LSP) for code intelligence. Use case: Swift/iOS development, SwiftUI autocomplete, Apple-platform code intelligence. (Claude Code)A - TypeScript/JavaScript language server for enhanced code intelligence. Use case: TS/JS autocomplete, type inference, refactor-rename across files. (Claude Code)A - Triage proposed AI use cases against your registry, run impact assessments across regimes in scope, review vendor AI terms, and keep AI policy current with practice. Use case: AI use-case triage, AI impact assessments, vendor AI-terms review. (Cowork)A - Review vendor agreements, NDAs, and SaaS subscriptions against your playbook, track renewals and cancel-by deadlines, and route escalations to the right approver. Use case: Vendor-contract review, renewal-deadline tracking, escalation routing. (Cowork)A - Run M&A diligence at scale with cited tabular review, build disclosure schedules and closing checklists, draft board consents, and track entity compliance across jurisdictions. Use case: M&A diligence review, disclosure-schedule building, board-consent drafting. (Cowork)A - Review hires and terminations for jurisdiction-specific risk, classify workers against the controlling state test, track leave deadlines, and draft policies with state supplements. Use case: Termination risk review, worker-classification tests, leave-deadline tracking. (Cowork)A - Run first-pass trademark clearance and freedom-to-operate triage, screen invention disclosures, draft cease-and-desist and DMCA letters, and track registration renewals. Use case: Trademark clearance, invention-disclosure screening, cease-and-desist drafting. (Cowork)A - Drill Socratically, brief cases, build outlines, run jurisdiction-tuned bar prep, and grade IRAC practice without writing the work for the student. Use case: Case briefing, outline building, bar-prep drilling and IRAC grading. (Cowork)A - Speed up contract review, NDA triage, and compliance workflows for in-house legal teams including legal-brief drafting and precedent research. Use case: Contract redlining, NDA triage, precedent research, compliance-workflow drafting. (Cowork)A - Find, evaluate, and install community legal skills with a security-review gate before anything lands in your environment. Use case: Legal-skill discovery, security-gated installation, community-skill evaluation. (Cowork)A - Set up a clinic, onboard students, run structured intake, track deadlines with malpractice-aware caution, and hand off cases at semester end within ABA Formal Op. 512. Use case: Clinic setup, structured client intake, deadline tracking and case handoff. (Cowork)A - Manage the litigation portfolio — matters, deadlines, holds, demands, and outside counsel — and produce claim charts, chronologies, depo prep, privilege logs, and brief drafts. Use case: Litigation-portfolio management, claim-chart building, deposition prep. (Cowork)A - Triage processing activities, generate PIAs, review DPAs as controller or processor, draft DSAR responses within statutory timelines, and monitor policy drift. Use case: PIA generation, DPA review, DSAR-response drafting. (Cowork)Community-maintained directory of plugins in Anthropic's official Claude Plugins catalog — 333 plugins (skills, MCP servers, slash commands, sub-agents, hooks) for Claude Code and Cowork. Not affiliat
概览
Community-maintained directory of plugins in Anthropic's official Claude Plugins catalog — 333 plugins (skills, MCP servers, slash commands, sub-agents, hooks) for Claude Code and Cowork. Not affiliated with Anthropic.
工具
暂无工具列表。远程服务器在首次健康检查后会获得工具列表。
常见问题
在本页复制适用于 Cursor、Claude Desktop、Claude Code 或 VS Code 的 stdio 配置。它使用发布者的安装方式,而不是 PlugMCP 代理。
此记录未列出必需的环境变量。
Awesome Claude Plugins 通过 stdio 在本地运行。
上次检查时间:2026-10-28。 Community-maintained directory of plugins in Anthropic's official Claude Plugins catalog — 333 plugins (skills, MCP servers, slash commands, sub-agents, hooks) for Claude Code and Cowork. Not affiliated with Anthropic.