TL;DR
Anthropic just shipped an official curated plugin directory for Claude Code. It earned 2,500+ stars in a single day and changes how you extend your AI coding workflow.
Read next
Ruflo crossed 37,700 GitHub stars this week, adding nearly 1,900 in a single day. It turns Claude Code into a coordinated swarm of 100+ specialized agents with MCP integration, distributed vector memory, and zero-trust agent federation.
7 min readA four-principle CLAUDE.md distilling Andrej Karpathy's observations on LLM coding mistakes gained 44k stars in a single week - and is now installable as a Claude Code plugin in two commands.
6 min readCodeGraph builds a local SQLite index of your codebase so Claude Code, Cursor, and Codex CLI spend far fewer tokens exploring files - trending on GitHub with 12k stars and real benchmark numbers.
6 min readanthropics/claude-plugins-official landed at number one on GitHub trending today with 2,556 new stars in 24 hours, pushing its total past 23,800. That velocity is not random. It is Anthropic's answer to a real pain point: developers have been shipping skills, hooks, MCP servers, and agents in scattered GitHub repos with no standard install path. The official plugin directory gives all of that a single, curated home inside Claude Code itself. When a first-party distribution channel ships from the tool's own creator, developers pay attention fast.
The claude-plugins-official repo is a curated marketplace of plugins that extend Claude Code's functionality. Think of it as the App Store for your AI coding environment - but open source, git-backed, and with a companion community tier for third-party submissions.
Plugins are structured directories that bundle any combination of:
The directory maintains two tiers. The claude-plugins-official side is curated by Anthropic and ships pre-configured in every Claude Code installation. A separate anthropics/claude-plugins-community repo holds third-party submissions that pass review. These install with a short @marketplace handle rather than a full git URL.
Each plugin follows a standard structure: a .claude-plugin/plugin.json manifest at the root for metadata, then skills/, agents/, hooks/, .mcp.json, and monitors/ directories alongside it. One important gotcha - all functional directories must sit at the plugin root, not inside .claude-plugin/. The docs flag that as the most common setup mistake.
If you are running a recent Claude Code build, the official marketplace is already registered. Browse available plugins with:
/plugin > Discover
Install a specific plugin directly:
/plugin install {plugin-name}@claude-plugins-official
To add the community marketplace and install from it:
/plugin marketplace add anthropics/claude-plugins-community
/plugin install {plugin-name}@claude-community
For local development, load a plugin directory without installing it:
claude --plugin-dir ./my-plugin
The --plugin-dir flag also accepts a .zip archive (requires Claude Code v2.1.128 or later):
claude --plugin-dir ./my-plugin.zip
During development, run /reload-plugins to pick up changes without restarting the session. Plugin skills are namespaced to prevent conflicts - a skill named hello in a plugin called my-plugin becomes /my-plugin:hello.
Get the weekly deep dive
Tutorials on Claude Code, AI agents, and dev tools - delivered free every week.
From the archive
May 22, 2026 • 8 min read
May 21, 2026 • 7 min read
May 20, 2026 • 7 min read
May 19, 2026 • 8 min read
Teams standardizing their Claude Code workflows - If your team has spread settings, hooks, and slash commands across individual .claude/ directories with no shared baseline, a plugin gives you one install command that propagates the full setup. The plugin's settings.json can activate a custom agent as the main thread when the plugin loads.
Developers who already built standalone skills - The migration path is documented. Copy commands/ and skills/ directories into a plugin root, create the plugin.json manifest, move hooks from settings.json into hooks/hooks.json, and test with --plugin-dir. The plugin version takes precedence over .claude/ configs with matching names.
Tool builders who want distribution - If you maintain a CLI or service that benefits from a dedicated Claude Code workflow (custom review commands, deploy hooks, context-aware diagnostics), packaging as a plugin gives you a standard install path. Approved community plugins are pinned to a commit SHA and synced nightly.
Solo developers who context-switch across many stacks - A per-project plugin that bundles LSP config, framework-specific agents, and relevant hooks can replace a growing pile of manual project setup.
This repo sits exactly where our community tooling lives. Developers Digest has been tracking the MCP ecosystem at mcp.developersdigest.tech, documenting skills patterns at skills.developersdigest.tech, and covering hooks workflows at hooks.developersdigest.tech. The plugin format unifies all three into one distributable unit.
The key shift is that skills, hooks, and MCP server configs no longer live in isolated repos with custom install docs. A single /plugin install command pulls the manifest and its bundled components together. That makes the patterns we have been documenting at DevDigest far more actionable - you can now point someone at a plugin URL and have them running your exact workflow in under a minute.
The bin/ directory support is especially relevant to the CLI-focused content at clis.developersdigest.tech. Plugins can ship executables that land directly in the Bash tool's PATH, which means CLI wrappers and helper scripts can ship alongside the skills that call them - no separate install steps required.
The directory structure is clean and the documentation at code.claude.com/docs/en/plugins is thorough. The namespacing approach (/plugin-name:skill-name) prevents conflicts between plugins sensibly.
The limitations are worth knowing upfront. The official marketplace (claude-plugins-official) is curated by Anthropic at its discretion with no application process - you cannot submit your plugin there. Community submissions go to claude-plugins-community and have a review pipeline plus a nightly sync delay, so approvals do not appear instantly. The security model is trust-based: Anthropic explicitly warns that it cannot verify MCP servers or bundled software in third-party plugins. Read the source before installing anything outside your own production.
The feature set is rich but some components - background monitors, --plugin-url remote loading - require specific Claude Code versions. Check your version before relying on those in team setups.
For the right use case (team standardization, shareable developer tooling), this is a meaningful quality-of-life upgrade over the current manual .claude/ copy-paste approach.
Technical content at the intersection of AI and development. Building with AI agents, Claude Code, and modern dev tools - then showing you exactly how it works.
Interactive TUI dashboard that shows exactly where your Claude Code and Cursor tokens are going, in real time.
View ToolAnthropic's agentic coding CLI. Runs in your terminal, edits files autonomously, spawns sub-agents, and maintains memory...
View ToolThe TypeScript toolkit for building AI apps. Unified API across OpenAI, Anthropic, Google. Streaming, tool calling, stru...
View ToolMCP server directory and ranking site. Tracks weekly downloads, GitHub stars, and build status across 5,000+ servers.
View ToolUnlock pro skills and share private collections with your team.
View AppEvery coding agent in one window. Stop alt-tabbing between Claude, Codex, and Cursor.
View AppTurn a one-liner into a working Claude Code skill. From idea to installed in a minute.
View AppLocal, project, user, and plugin-level MCP configurations.
Claude CodeThe primary command-line entry point for Claude Code sessions.
Claude CodeReal-time prompt loop with history, completions, and multiline input.
Claude Code
Nimbalyst Demo: A Visual Workspace for Codex + Claude Code with Kanban, Plans, and AI Commits Try it: https://nimbalyst.com/ Star Repo Here: https://github.com/Nimbalyst/nimbalyst This video demos N...

Composio: Connect AI Agents to 1,000+ Apps via CLI (Gmail, Google Docs/Sheets, Hacker News Workflows) Check out Composio here: http://dashboard.composio.dev/?utm_source=Youtube&utm_channel=0426&utm_...

Anthropic has released Channels for Claude Code, enabling external events (CI alerts, production errors, PR comments, Discord/Telegram messages, webhooks, cron jobs, logs, and monitoring signals) to b...
CodeGraph builds a local SQLite index of your codebase so Claude Code, Cursor, and Codex CLI spend far fewer tokens expl...
CodeGraph hit 7,800+ stars with 1,900 added in a single day - a local MCP knowledge graph that lets Claude Code explore...
agentmemory is a self-hosted MCP server that gives Claude Code, Cursor, and Gemini CLI searchable long-term memory acros...
agentmemory gives AI coding agents a persistent brain - capturing session context automatically via 12 Claude Code hooks...
Ruflo is an open-source multi-agent orchestration platform built specifically for Claude, shipping 100+ specialized agen...
TypeScript educator Matt Pocock open-sourced his personal Claude Code skills, and 60,000 developers starred the repo in...

New tutorials, open-source projects, and deep dives on coding agents - delivered weekly.