# MCPBundles | Give Your AI Superpowers Agent operations (directory, WebMCP, connect paths): https://www.mcpbundles.com/agents.md Directory API v1 (headless search + invoke): https://www.mcpbundles.com/api/directory/v1 — OpenAPI at `/openapi.json`; prefer over registry for fetch agents. MCP Registry API (VS Code gallery shape): https://www.mcpbundles.com/api/registry — `chat.mcp.gallery.serviceUrl` = that base URL; read-only metadata. Directory API and Registry API limits and license: free is 30 requests/minute (2/second) per IP, no account. Directory API Pro is $299/month for 300 requests/minute (10/second) per workspace, using the workspace API key as `Authorization: Bearer`; a human subscribes at https://www.mcpbundles.com/pricing#directory-api. On `429`, honor `Retry-After`; if the body says `kind: "upgrade_required"`, show the user `upgrade_url` instead of retrying in a loop. Using directory data in a product you run for customers needs Pro. On every tier, do not resell or republish directory data as a dataset, feed, API, or competing index: https://www.mcpbundles.com/terms WebMCP manifest: https://www.mcpbundles.com/.well-known/webmcp Markdown for agents (HTML www pages): send `Accept: text/markdown` or append `.md` to the path (e.g. https://www.mcpbundles.com/pricing.md, https://www.mcpbundles.com/index.md). Do not send `Accept: text/markdown` for this file or https://www.mcpbundles.com/agents.md — fetch them as plain GET. Directory search markdown: https://www.mcpbundles.com/?q=stripe with `Accept: text/markdown`. MCPBundles is a cloud platform that connects AI assistants to real-world tools and APIs through the Model Context Protocol (MCP). Enable ChatGPT, Claude, Cursor, VS Code, and other MCP-compatible AI tools to perform real actions—manage campaigns, update CRMs, deploy code, send notifications, and hundreds more. ## Product - Product: Cloud-hosted MCP server platform - URL: https://www.mcpbundles.com - Primary Use Case: Connect AI assistants to external APIs and tools - Key Features: Browse the MCP directory, connect hosted and remote MCP servers, secure credential management, OAuth and API key support, credential verification, interactive tool testing via Studio, programmatic tool calling via Hub ## Supported AI Platforms ChatGPT, Claude Desktop, Claude Code, Cursor, VS Code, Raycast, Windsurf, Gemini CLI, Gemini Code, Goose, Devin, and any MCP-compatible client. ## Key Concepts - MCP server: A hosted collection of related tools exposed via a single MCP URL (e.g., a GitHub MCP server with repo, issue, and PR tools) - Tool: An individual action an AI can perform (e.g., list_campaigns, create_contact, send_message) - Integration: An external service an MCP server connects to (e.g., Smartlead, GitHub, Slack, HubSpot) - Credential: API keys or OAuth tokens that authenticate with providers - MCP URL: The connection string added to AI clients (format: https://mcp.mcpbundles.com/bundle/slug) - Studio: Interactive browser-based workspace to test tools before deploying to AI - The Hub: Cross-service workflow orchestration using programmatic tool calling ## Categories Analytics & Insights, Backend Services, Cloud Services, Collaboration, Customer Relations, Developer Tools, E-commerce, Education, Email Services, Enterprise Software, Entertainment, Financial Services, Forms & Surveys, Web Hosting, IoT, Location Services, Machine Learning, Marketing & Sales, Media & Content, Messaging & Communication, Monitoring & Logging, Open Data, Payment Processing, Project Management, Search & Discovery, Security & Auth, Social Media, Storage & Files, Customer Support, Telecommunications, Text Processing, Time Management, Utilities & Tools, Transportation ## Featured Integrations Popular MCP servers: GitHub, Slack, HubSpot, Notion, Stripe, PostgreSQL, Ahrefs, Google Analytics, Sentry, Resend, PostHog, Trello, Asana, Linear, Zendesk, Salesforce, Mailchimp, Intercom, Twilio, and more. ## Example Use Cases Marketing: "Check my Smartlead campaigns and create a performance report" Development: "List my open GitHub PRs and summarize each one" CRM: "Add these 5 contacts to HubSpot with their company info" Communication: "Send a Slack message to #marketing with today's metrics" Analytics: "Pull my PostHog funnel data and identify drop-off points" SEO: "Get my top performing pages from Ahrefs and suggest improvements" ## Documentation Full documentation: https://mcpbundles.com/docs Quick Start Guide: https://mcpbundles.com/docs/getting-started What is MCP?: https://mcpbundles.com/docs/what-is-mcp Integration Guides: https://mcpbundles.com/docs/integrations/chatgpt FAQ: https://mcpbundles.com/docs/faq ## Key Pages - Homepage / MCP Directory: https://www.mcpbundles.com - MCP Registry API (directory subregistry): https://www.mcpbundles.com/api/registry - MCP server detail (catalog): https://www.mcpbundles.com/skills/{slug} - MCP Apps browse facet: https://www.mcpbundles.com/?browse=1&app=1 - Studio: https://www.mcpbundles.com/studio - Pricing: https://www.mcpbundles.com/pricing - Learning Hub: https://www.mcpbundles.com/learn - MCP Security Methodology (live rule catalog with mcp-scan + SAFE-MCP citations): https://www.mcpbundles.com/learn/mcp-security - Supported Platforms: https://www.mcpbundles.com/what-is-mcp/platforms - What is an MCP Server?: https://www.mcpbundles.com/what-is-mcp-server - How It Works: https://www.mcpbundles.com/how-it-works ## Getting Started 1. Sign up at https://www.mcpbundles.com 2. Browse the directory at https://www.mcpbundles.com and pick an MCP server 3. Add credentials for the server's providers (OAuth or API key) 4. Enable the MCP server in your workspace 5. Use it in Studio or add the MCP server URL to your AI client 6. Chat to the AI and use the tools ## Technical Details - Protocol: Model Context Protocol (MCP) - open standard - Transport: HTTP Streamable and SSE for hosted servers - Authentication: OAuth 2.0 and API key support per provider - Credential Storage: Encrypted at rest, never exposed to AI models - .mcpb: Open MCP packaging format (ZIP + manifest) used by some clients such as Claude Desktop; MCPBundles servers connect via MCP URL