Skip to content
Back to search

tomcp

Turn any website or doc into an MCP server

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/ami3466--tomcp/badge.svg)](/listings/ami3466--tomcp)

HTML

<a href="/listings/ami3466--tomcp"><img src="/api/listings/ami3466--tomcp/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Put `tomcp.org/` in front of any URL:

https://tomcp.org/docs.stripe.com

That address is an MCP server. Add it to your client, or paste the URL at [tomcp.org](https://tomcp.org) instead.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 164 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # toMCP
  2. **Turn any website into an MCP server**
  3. ## Quick start
  4. Put `tomcp.org/` in front of any URL:
  5. ```
  6. https://tomcp.org/docs.stripe.com
  7. ```
  8. That address is an MCP server. Add it to your client, or paste the URL at [tomcp.org](https://tomcp.org) instead.
  9. ## Why toMCP?
  10. specific APIs often cause AI hallucinations, and web searching is unreliable. This tool lets you serve your documentation as an MCP server, giving the AI direct access to the clear context it needs without manual copy-pasting.
  11. Standard `web_fetch` tools dump raw HTML into your AI's context—navbars, scripts, footers, and noise. toMCP runs pages through a **readability parser** and converts to **clean markdown**, using a fraction of the tokens.
  12. ### Persistent Documentation Context
  13. AI assistants hallucinate API details when they lack documentation. MCP Resources are **pinned as permanent, read-only context**—the model won't skip or forget them. Ideal for framework docs, API references, and internal team docs.
  14. ### web_fetch vs MCP Resources
  15. | | web_fetch | MCP Resource |
  16. |--|-----------|--------------|
  17. | Data | Raw HTML with noise | Clean markdown |
  18. | Tokens | High | Low |
  19. | Persistence | Per-request | Always available |
  20. | Hallucination | Higher | Lower |
  21. | JS Support | Full (SPAs / Dynamic) | Static Only (SSG) |
  22. ## Demo
  23. [![toMCP Demo](https://img.youtube.com/vi/-o2_T8TB9dQ/maxresdefault.jpg)](https://www.youtube.com/watch?v=-o2_T8TB9dQ)
  24. ## Supported AI Tools
  25. - **Cursor** - `~/.cursor/mcp.json`
  26. - **Claude Desktop** - `~/.claude/claude_desktop_config.json`
  27. - **Windsurf** - `~/.codeium/windsurf/mcp_config.json`
  28. - **VS Code** - `.vscode/mcp.json`
  29. - **Cline** - `~/.cline/mcp_settings.json`
  30. ## How It Works
  31. ### MCP Config
  32. 1. Visit [tomcp.org](https://tomcp.org)
  33. 2. Enter any website URL
  34. 3. Select your AI tool
  35. 4. Copy the generated MCP config
  36. 5. Add it to your tool's config file
  37. 6. Restart your AI tool
  38. ### Chat
  39. 1. Visit [tomcp.org](https://tomcp.org)
  40. 2. Paste any website URL
  41. 3. Click "Start Chat"
  42. 4. Ask questions about the website's content
  43. ## Example Config
  44. ```json
  45. {
  46. "mcpServers": {
  47. "docs-stripe-com": {
  48. "url": "https://tomcp.org/docs.stripe.com"
  49. }
  50. }
  51. }
  52. ```
  53. ## Chat API
  54. ```bash
  55. curl -X POST https://tomcp.org/chat \
  56. -H "Content-Type: application/json" \
  57. -d '{"url": "docs.stripe.com", "message": "How do I create a payment intent?"}'
  58. ```
  59. ## AI Models
  60. ### Free Models (No API Key Required)
  61. These models are available for everyone with no setup:
  62. - **Llama 3.1 8B** (Meta) - Default model, fast and capable
  63. - **Hermes 2 Pro** (NousResearch) - Great for reasoning
  64. - **Mistral 7B** (Mistral) - Efficient instruction-following
  65. - **Gemma 7B LoRA** (Google) - Lightweight and fast
  66. ### paid Models (API Key Required)
  67. Add your Cloudflare Workers AI API key to unlock these models:
  68. - **Llama 3.3 70B** (Meta) - Most powerful Llama model
  69. - **DeepSeek R1 32B** (DeepSeek) - Advanced reasoning
  70. - **Mistral Large** (Mistral) - Enterprise-grade
  71. - **Gemma 3 12B** (Google) - Latest Gemma
  72. - **GPT OSS 120B/20B** (OpenAI) - Open-source GPT variants
  73. ## Adding Your API Key
  74. You can add your own Cloudflare Workers AI API key to:
  75. 1. **Unlock all paid models** - Access larger, more capable models
  76. 2. **Bypass rate limits** - No daily request limits
  77. 3. **Use your own quota** - Charges go to your Cloudflare account
  78. ### How to Get an API Key
  79. 1. Go to [Cloudflare Workers AI](https://developers.cloudflare.com/workers-ai/get-started/rest-api/#1-get-api-token-and-account-id)
  80. 2. Create an API token with Workers AI permissions
  81. 3. Copy the token
  82. ### How to Add Your Key
  83. 1. Start a chat session on [tomcp.org](https://tomcp.org)
  84. 2. Below the chat input, you'll see "Add API key from Cloudflare Workers AI"
  85. 3. Paste your API key and click "Save"
  86. 4. paid models will now be unlocked in the dropdown
  87. ### Where Is the API Key Stored?
  88. - Your API key is stored **locally in your browser** using `localStorage`
  89. - Key name: `tomcp_api_key`
  90. - The key is sent with each chat request but **never stored on our servers**
  91. - You can remove it anytime by clicking "Remove" in the API key section
  92. ## How It Works (Technical)
  93. ### Model Fetching
  94. The available models are fetched dynamically from the Cloudflare Workers AI API:
  95. 1. Frontend calls `GET /models` endpoint on page load
  96. 2. Worker fetches models from `api.cloudflare.com/client/v4/accounts/{id}/ai/models/search`
  97. 3. Models are filtered to "Text Generation" tasks and cached for 5 minutes
  98. 4. Frontend displays free models as enabled, paid models as disabled (until API key is added)
  99. ### Chat Flow
  100. 1. User enters a URL and starts chatting
  101. 2. Worker fetches the static HTML and converts it to clean Markdown (JavaScript is not executed, so SPAs or dynamically-loaded content won't be captured)
  102. 3. Content is sent to the selected AI model with the user's message
  103. 4. Response is returned to the user
  104. ### Rate Limiting
  105. Without an API key:
  106. - 5 requests per IP per day
  107. With your API key:
  108. - No rate limits (uses your Cloudflare account quota)
  109. ## Tech Stack
  110. - **Frontend**: Vanilla HTML/CSS/JS with Tailwind CSS
  111. - **Backend**: Cloudflare Workers
  112. - **AI**: Cloudflare Workers AI (multiple models)
  113. ## Features
  114. - Works with any public URL
  115. - No setup required - just paste the config
  116. - Free forever - powered by Cloudflare Workers
  117. - Chat with any website using AI
  118. - Side-by-side MCP Config + Chat interface
  119. - **Multiple AI models** - Choose from Llama, Mistral, Gemma, and more
  120. - **Bring your own API key** - Unlock paid models and bypass rate limits
  121. ## License
  122. Apache 2.0

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.