Browse by category
MCP Servers
Model Context Protocol servers — give Claude Code structured tool access to filesystems, APIs, databases, browsers, and more.
7 toolsAI Coding Agents
Cursor, Windsurf, Aider, Cline, Continue, GitHub Copilot — the IDE-and-terminal agents that complement Claude Code.
6 toolsAI Providers
Anthropic, OpenAI, Google Gemini, xAI, Groq, DeepSeek — the LLM backends Claude Code can route to via env keys.
3 toolsLLM Routers
LiteLLM, OpenRouter, Portkey — single-API multi-provider proxies with caching, fallbacks, and budget controls.
4 toolsEval & Observability
Langfuse, LangSmith, Helicone, Braintrust — trace, eval, and monitor your LLM apps in production.
6 toolsVector Databases
Pinecone, Qdrant, Weaviate, Chroma, pgvector, LanceDB — embedding storage and similarity search for RAG.
5 toolsAgent Frameworks
LangChain, LlamaIndex, CrewAI, AutoGen, PydanticAI — Python/JS frameworks for building multi-step LLM apps.
3 toolsSandboxes & Execution
E2B, Modal, Daytona — secure cloud environments for running AI-generated code or persistent dev workspaces.
734 toolsVS Code Extensions
Continue, Cody, Cline, Tabnine, Codeium — IDE extensions for AI-assisted coding inside VS Code.
92 toolsRules Libraries
Cursor-rules and Claude-Code rule packs — curated instructions that improve agent behavior for a given language or stack.
Arcade Pro
The free Arcade catalog above is unrated and alphabetic — pick what you need, browse without an account. Arcade Pro adds three things that turn discovery into decision:
- Quality Score on every entity (0–100, content-derived from 4 signals: structural completeness, content richness, install actionability, license clarity) — sorts the catalog by what's actually shippable, not what got the most stars.
- 1-click install from the desktop app — every entity carries its install_type (mcp_npm, pip_package, vscode_ext, etc.) so the desktop installer renders the right one-button action.
- Stack Diagnostics — point the desktop app at your project and get a recommendation: which MCP servers you're missing, which agent framework matches your codebase, what's overdue for an upgrade.
Arcade Monthly
$5/mo
or $49/year · or $99 lifetime
Quality Score on every entity
1-click install (desktop app)
Stack Diagnostics
Coming soonin QA · v1.0.18
Bundle (Pro + Arcade)
$14/mo
or $128/year · or $248 lifetime
Everything in Arcade
Everything in ClaudSkills Pro (Quality Score on 70k+ skills + 1-click install)
One renewal date, one invoice
Coming soonin QA · v1.0.18
How discovery works
The Arcade miner runs daily at 02:30 local. It hits seven public sources in sequence:
- Official MCP registry (
registry.modelcontextprotocol.io) — the authoritative source. - Awesome-MCP repos — curated GitHub README lists with hand-checked entries.
- GitHub topic search —
topic:mcp stars:>=10for breadth. - npm —
@modelcontextprotocol/*andmcp-server-*packages. - PyPI — known-prefix probe (no public search API).
- VS Code Marketplace — AI / MCP / Claude-tagged extensions with ≥100 installs.
- GitHub
topic:cursor-rules— rule packs that work with Cursor, Cline, and Claude Code.
Cross-source deduplication is by slug — the same MCP server discovered four ways collapses into one record with all sources tracked. Quality scoring runs after every mine cycle.
Open data
The full Arcade catalog (without the Pro Quality Score field) is at /data/arcade.json. CC BY 4.0 — free for any use including commercial, attribution required. The Quality Score is a content-derived ranking we compute in-house; it's Pro-gated to fund the catalog's daily maintenance.