Generation is one call; what shapes the result is everything you set before it — outline, style guide, work description, details and vision are separate fields, not one prompt blob. That separation is the reason to drive this from a conversation, where you can revise the style guide and regenerate without rebuilding the whole brief. Every tool also takes the API key as an argument, so check how your client stores tool inputs.
The official MCP server for the DeepWriter API. It exposes project management and content generation as MCP tools, so an assistant can set up a writing project with its outline, style guide and vision, then trigger generation against it.
- Every project on your account listed — `listProjects`; one project's detail — `getProjectDetails`
- A new project created from a title and email — `createProject`
- A project updated across the fields that steer output — `title`, `prompt`, `author`, `model`, `outline_text`, `style_text`, `supplemental_info`, `work_description`, `work_details`, `work_vision` — `updateProject`
- Content generated for a project using DeepWriter's AI — `generateWork`, with an optional `is_default` that defaults to true
- A project deleted — `deleteProject`
- MCP protocol version 2025-03-26 over stdio transport, with structured logging at configurable levels
A DeepWriter API key, set as `DEEPWRITER_API_KEY` in the client's env block or a `.env` file. Node.js v17 or higher and npm v6 or higher. Install through Smithery with `npx -y @smithery/cli install @deepwriter-ai/deepwriter-mcp --client claude`, or clone, `npm install`, `npm run build` and run `node build/index.js`.
One command plus a key — npx -y @smithery/cli install @deepwriter-ai/deepwriter-mcp --client claude, then supply credentials