Prompt Optimizer MCP
An MCP server and web tool for improving prompts: three tools for optimizing system and user prompts with iterative refinement
- 41,320 ★ on the repository
- Last commit
Extended access
Why this level
- It sends prompt text to the configured LLM provider for optimization
Install
- Works in
Manual install
docker run -d -p 8081:80 -e VITE_OPENAI_API_KEY=your-openai-key -e MCP_DEFAULT_MODEL_PROVIDER=openai --name prompt-optimizer linshen/prompt-optimizerStarts the web UI and the MCP server together; MCP is available at /mcp.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- It sends prompt text to the configured LLM provider for optimization
This is third-party code. Review the repository files before installing.
What it does
Prompt Optimizer takes a draft prompt and rewrites it so the model gives a more precise, predictable answer. Through MCP it exposes three tools: optimizing a user prompt, optimizing a system prompt, and iteratively refining an already-solid prompt against a specific requirement. Besides the MCP server, the project has a web app, a desktop app and a Chrome extension with the same optimization logic, testing, result evaluation and a library of saved prompts.
Who it is for. For anyone who writes LLM prompts regularly and wants better, more stable wording without manually trying variants.
Good fit when
- You need to improve a system or user prompt right from an agent conversation
- The prompt already works but needs targeted refinement for a new requirement
- You need a prompt library with version history and tests
Not a fit when
- The prompt is one-off and simple, so going through MCP is overkill
- No LLM provider API key is configured for the optimizer itself
Example request
Optimize this system prompt via prompt-optimizer so the model stays on topic more reliablyMCP
- Transport
- http
- Authentication
- not required
Environment variables
VITE_OPENAI_API_KEYsecret- One of the supported LLM provider keys; at least one is required.
MCP_DEFAULT_MODEL_PROVIDERrequired- The default provider used for optimization requests.
Limitations
The optimizer itself calls a separate LLM, so at least one provider API key is required (OpenAI, Gemini, DeepSeek, Zhipu and others, including local ones via Ollama). The MCP server is not published as a standalone npm package; it is installed via the Docker image or built from the monorepo source. Licensed under AGPL-3.0.
How to disable. Stop and remove the container: docker stop prompt-optimizer && docker rm prompt-optimizer, and remove the MCP server from the agent config.
FAQ
Is Docker mandatory?
No, the README also describes a local pnpm build for developers, but Docker is the recommended path for regular users.
Which provider does the optimizer itself need?
At least one supported key: OpenAI, Gemini, DeepSeek, Grok, SiliconFlow, Zhipu, or a custom endpoint such as Ollama.
Discussion
No questions or reviews yet.
Related
MCP serversOpen prompt library with a Claude Code plugin, MCP server and CLI: search, fetch and improve prompts and skills from an agent
A prompt template store with search and variable filling, plus a script that runs Claude agents across a whole project.
MCP server for prompt templates, step chains and output validation, exported as skills for Claude Code, Cursor and Codex
A skill that writes precise prompts tailored to a specific tool: Cursor, Claude Code, Midjourney, Stable Diffusion and more
An MCP server that inserts a human feedback step into agent tasks: a web or desktop window for text, images and prompts
PromptOps keeps Claude Code, Codex and Gemini CLI side by side by project and gives a lead agent MCP to launch helpers
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Prompt Optimizer (this card) | yes | not required | 41,320 | |
| WuMing Skills | no | 752 | a Chinese-language skill set for Claude Code | |
| Prompt Architect | no | 333 | a skill that scores a prompt on five criteria |
README in short
The README describes Prompt Optimizer as a prompt-optimization tool with four usage paths: web app, desktop, Chrome extension and Docker deployment. A separate MCP server document lists three tools: optimize-user-prompt, optimize-system-prompt and iterate-prompt, running over Streamable HTTP, and shows connecting to Claude Desktop via a services.json file.