system-prompts-mcp-server
About
Model Context Protocol server exposing system prompt files and summaries.
Details
- Author
- jamesanz
- Categories
- Productivity, Other, Knowledge Base
Jump to
Setup
Install system-prompts-mcp-server in your MCP client (Claude Desktop, Cursor, Windsurf, and others).
Repository: https://github.com/jamesanz/system-prompts-mcp-server
Follow the installation instructions in the repository README, then restart your MCP client.
Access system prompts from AI tools in your workflow.Browse and fetch prompts from Devin, Cursor, Claude, GPT, and more. Model-aware suggestions help you find the perfect prompt for your LLM.
- πAutomatic Discoveryβ Every prompt inprompts/is automatically exposed as an MCP tool
- π―Model-Aware Suggestionsβ Get prompt recommendations based on your LLM (Claude, GPT, Gemini, etc.)
- πComprehensive Collectionβ Access prompts from Devin, Cursor, Claude, GPT, and more
- πEasy Setupβ One-click install in Cursor or simple manual setup
- π§Extensibleβ Add your own prompts and they're automatically available
Ready to explore system prompts? Install in seconds:
npm install -g system-prompts-mcp-server # Or from source: git clone https://github.com/JamesANZ/system-prompts-and-models-of-ai-tools.git cd system-prompts-and-models-of-ai-tools && npm install && npm run build
- list_promptsβ Browse available prompts with filters (service, flavor, provider)
- get_prompt_suggestionβ Get ranked prompt suggestions for your LLM and keywords
- <service>-<variant>-<flavor>β Direct access to any prompt (e.g.,cursor-agent-system,devin-summary)
- Scansprompts/directory for.txt,.md,.yaml,.yml,.jsonfiles
- Each file becomes a dedicated MCP tool
- Infers metadata (service, variant, LLM family, persona hints)
- Each tool call includes a reminder for the model to embody the loaded prompt
- Helps models behave like the original service (Devin, Cursor, etc.)
cursor://anysphere.cursor-deeplink/mcp/install?name=system-prompts-mcp&config=eyJzeXN0ZW0tcHJvbXB0cy1tY3AiOnsiY29tbWFuZCI6Im5weCIsImFyZ3MiOlsiLXkiLCJzeXN0ZW0tcHJvbXB0cy1tY3Atc2VydmVyIl19fQ==
# Clone and build git clone https://github.com/JamesANZ/system-prompts-and-models-of-ai-tools.git cd system-prompts-and-models-of-ai-tools npm install npm run build # Run server npm start
macOS:~/Library/Application Support/Claude/claude_desktop_config.json
Windows:%APPDATA%\Claude\claude_desktop_config.json
{ "mcpServers": { "system-prompts-mcp": { "command": "node", "args": ["/absolute/path/to/system-prompts-and-models-of-ai-tools/dist/index.js"], "env": { "PROMPT_LIBRARY_ROOT": "/absolute/path/to/system-prompts-and-models-of-ai-tools/prompts" } } } }
Restart Claude Desktop after configuration.
{ "tool": "list_prompts", "arguments": { "service": "cursor", "flavor": "system", "limit": 10 } }
Find the best prompt for your LLM and use case:
{ "tool": "get_prompt_suggestion", "arguments": { "userLlm": "claude-3.5-sonnet", "keywords": ["code", "pair programming"] } }
Call a prompt directly by its tool name:
{ "tool": "cursor-agent-system", "arguments": {} }
{ "tool": "cursor-agent-system", "arguments": { "format": "json" } }
Add prompts by placing files in theprompts/directory:
Supported formats:.txt,.md,.yaml,.yml,.json
prompts/My Service/ βββ System Prompt.txt β Tool: "my-service-system-prompt-system" βββ tools.json β Tool: "my-service-tools-tools"
- Directory names become the service name
- File names create tool variants
- Files are automatically classified as system prompts, tools, or summaries
After adding prompts, restart the MCP server. Uselist_promptsto find your custom prompts.
Custom directory:SetPROMPT_LIBRARY_ROOTenvironment variable to use a different location.
- AI Tool Developersβ Reference and adapt prompts from successful AI tools
- Researchersβ Study how different tools structure their system prompts
- Developersβ Find the perfect prompt for your LLM and use case
- Prompt Engineersβ Compare and learn from proven prompt patterns
Built with:Node.js, TypeScript, MCP SDK
Dependencies:@modelcontextprotocol/sdk,zod
Platforms:macOS, Windows, Linux
- PROMPT_LIBRARY_ROOT(optional): Override prompt root directory (defaults toprompts/)
- src/β TypeScript MCP server implementation
- dist/β Compiled JavaScript
- prompts/β Prompt library and original documentation
βIf this project helps you, please star it on GitHub!β
Contributions welcome! Feel free to adapt the discovery logic, add tests, or extend metadata inference for new prompt formats.
See the original repository for license information.
If you find this project useful, consider supporting it:
lnbc1pjhhsqepp5mjgwnvg0z53shm22hfe9us289lnaqkwv8rn2s0rtekg5vvj56xnqdqqcqzzsxqyz5vqsp5gu6vh9hyp94c7t3tkpqrp2r059t4vrw7ps78a4n0a2u52678c7yq9qyyssq7zcferywka50wcy75skjfrdrk930cuyx24rg55cwfuzxs49rc9c53mpz6zug5y2544pt8y9jflnq0ltlha26ed846jh0y7n4gm8jd3qqaautqa
βΏ Bitcoin:bc1ptzvr93pn959xq4et6sqzpfnkk2args22ewv5u2th4ps7hshfaqrshe0xtp
Ξ Ethereum/EVM:0x42ea529282DDE0AA87B42d9E83316eb23FE62c3f
Official Notion MCP server for searching, reading, creating, and updating Notion pages, databases, and workspace content from AI agents.
MCP server for Apple Notes with semantic search and CRUD operations. Claude searches, reads, creates, updates, and manages your Apple Notes through natural language.
Markdown-first Notion MCP server with 9 composite tools, 39 actions, and ~77% token reduction via tiered docs.
AI-powered reading triage MCP. 26 tools with a 24h burn timer. Read less, absorb more.
Integrates Model Context Protocol (MCP) with Obsidian, allowing AI assistants to interact with your notes and vault.
An MCP server that enables AI assistants to interact with Confluence content through a standardized interface.
An MCP server for document ingestion, chunking, semantic search, and note management.
Remote MCP server for Docsie video-to-docs automation and workspace knowledge search.
An AI-powered documentation management system with hierarchical context management and session continuity.
Search, cite, and read PDFs from your EndNote reference library directly in Claude Desktop
Sign in to leave a review
Use Google, GitHub, or an email account so ratings stay tied to real people.
No reviews posted yet.





