MCP | ReplyCap

What ReplyCap MCP gives you

A lightweight MCP layer that helps AI clients prepare structured screenshot data and route users to the right ReplyCap editor.

Right tool routing
Match each request to the correct ReplyCap generator faster.

Structured inputs
Return clean payloads for X tweet, classic Twitter tweet, iMessage, TikTok, and Shorts setups.

Browser export
Keep final screenshot rendering simple in the existing web editor.

Available MCP tools

Each tool maps to one of the current ReplyCap generators.

Tool	Purpose	Editor
`list_replycap_generators`	List all ReplyCap editors	Open editor
`create_x_tweet_screenshot`	Prepare X / Twitter screenshot data	Open editor
`create_twitter_tweet_screenshot`	Prepare classic Twitter tweet data	Open editor
`create_imessage_screenshot`	Prepare iPhone iMessage conversation data	Open editor
`create_tiktok_comment_screenshot`	Prepare TikTok comment screenshot data	Open editor
`create_youtube_shorts_comment_screenshot`	Prepare YouTube Shorts comment data	Open editor

Quick start

Recommended flow for local MCP usage.

Clone the repo and install dependencies with pnpm.
Run `pnpm mcp` from the project root.
Add the server to your MCP client config using an absolute path.
Ask your client to call one of the ReplyCap tools to prepare screenshot data.
Open the returned ReplyCap editor URL and export the final image in the browser.

pnpm mcp

pnpm mcp

Claude Desktop example

Add an absolute project path, save the config, then restart the app.

{
  "mcpServers": {
    "replycap": {
      "command": "pnpm",
      "args": [
        "--dir",
        "/ABSOLUTE/PATH/TO/replycap",
        "mcp"
      ]
    }
  }
}

{
  "mcpServers": {
    "replycap": {
      "command": "pnpm",
      "args": [
        "--dir",
        "/ABSOLUTE/PATH/TO/replycap",
        "mcp"
      ]
    }
  }
}

Example prompts

You can paste these into an MCP-aware client after connecting the server.

Create an X tweet screenshot for a product launch post.
Prepare a classic Twitter tweet screenshot with Twitter Web App as the source.
Generate an iMessage conversation between two friends planning dinner.
Prepare a TikTok comment screenshot with a dark theme and high likes.
Make a YouTube Shorts comment screenshot with 65 replies and a light theme.

What the server returns

Each MCP tool returns structured values plus the matching ReplyCap editor URL.

Suggested field values for the selected generator
The correct editor URL on ReplyCap
A clean payload that can be reused or adjusted by the client
Guidance for the final browser-based export step

Notes and limitations

The current MCP release is intentionally simple and stable.

The ReplyCap MCP server runs over stdio, which is the recommended local MCP pattern.
Final PNG rendering is still done in the website editor, not by the server itself.
This keeps the first version lightweight and aligned with the current browser-based preview system.