This server allows you to manage and improve prompts using the Pretty Prompt library and prompt improver.
List, save, and manage prompts in your personal library.
Organize your prompts into folders.
Improve and refine prompts using Pretty Prompt's AI capabilities, consuming your prompt credits.
Integrate with MCP clients like Cursor and Claude Desktop.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Pretty Prompt MCP Server list my saved prompts"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP server that exposes your Pretty Prompt library and prompt improver to Cursor, Claude Desktop, and other MCP clients.
Tool | Description |
| List saved prompts from your library |
| Fetch one prompt by id (full text, folder, tags, notes) |
| Save a prompt to your library |
| Update title, text, favourite status, or notes |
| Favourite or unfavourite a prompt by id |
| Permanently delete a prompt from your library |
| List folders in your library |
| Create a folder (optionally nested under a parent) |
| Rename an existing folder |
| Delete a folder (prompts move to root; optional strategies) |
| Move a prompt into a folder (or to root) |
| List all tags (UUID id, name, color) |
| Fetch one tag by UUID |
| Create a new tag |
| Rename or recolor a tag by UUID |
| Delete a tag by UUID |
| Apply a tag to a prompt by ids |
| Remove a tag from a prompt by ids |
| List saved context snippets (numeric ids for improve_prompt) |
| Create a context snippet (optional personal/work tag) |
| Update a context snippet's title, text, or global flag |
| Permanently delete a context snippet by id |
| Improve/refine a prompt (uses your Pretty Prompt credits) |
Related MCP server: MCP Prompt Manager
Go to pretty-prompt.com/settings/mcp and create an API key.
npm install
npm run buildAdd to your Cursor MCP config (~/.cursor/mcp.json or project
.cursor/mcp.json):
{
"mcpServers": {
"pretty-prompt": {
"command": "npx",
"args": ["-y", "@pretty-prompt/mcp@latest"],
"env": {
"PRETTY_PROMPT_API_KEY": "pp_mcp_..."
}
}
}
}For local development, override the production defaults:
"env": {
"PRETTY_PROMPT_API_KEY": "pp_mcp_...",
"PRETTY_PROMPT_BACKEND_URL": "http://0.0.0.0:8000",
"PRETTY_PROMPT_SUPABASE_URL": "http://127.0.0.1:54321",
"PRETTY_PROMPT_SUPABASE_ANON_KEY": "your-local-anon-key"
}Variable | Required | Default | Description |
| Yes | — | API key from settings ( |
| No |
| FastAPI backend URL |
| No |
| Supabase project URL |
| No | Production anon key | Supabase anon key |
| No | enabled | Set to |
| No | Production PostHog key | Override PostHog project key |
The MCP server sends PostHog events (fire-and-forget) for:
Event | When |
| Each time the server process starts and authenticates |
| First successful connection for an API key on this machine |
| Each tool invocation (name, success, duration, sanitized params) |
Events are attributed to the user's email from the MCP access token. Prompt text is never sent — only lengths and flags.
Disable locally with PRETTY_PROMPT_ANALYTICS=false.
Override env vars to point at local services:
PRETTY_PROMPT_BACKEND_URL=http://0.0.0.0:8000
PRETTY_PROMPT_SUPABASE_URL=http://127.0.0.1:54321Ensure the backend has SUPABASE_JWT_SECRET set and the mcp_api_keys
migration is applied.
Improve and refine a prompt using Pretty Prompt. Returns a polished prompt (flattened text plus structured sections), an explanation, and optional follow-up questions. Uses the user's prompt credits. To refine further, call again with question_answers from the follow_up_questions.
Inputs: prompt (required), category, provider, save_to_library, +3 more.
List folders in the user's Pretty Prompt library. Use this before move_prompt_to_folder to get folder UUIDs. Folders are a flat list; parent_id indicates nesting (null = root).
Inputs: no named parameters.
List saved prompts from the user's Pretty Prompt library.
Inputs: limit, offset, search, favorites_only.
Move a library prompt into a folder, or to the library root. Call list_library_folders first to resolve folder UUIDs. Pass folder_id=null to remove the prompt from its folder.
Inputs: prompt_id (required), folder_id.
Save a prompt to the user's Pretty Prompt library.
Inputs: prompt (required), title.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 5 tools
Each tool targets a distinct resource and action: listing prompts, saving prompts, listing folders, moving prompts, and improving prompts. There is no overlap or ambiguity between them.
Most tools follow a clear verb_noun or verb_preposition_noun pattern (list_library_prompts, save_to_library, move_prompt_to_folder), but the mix of styles (list_*, save_to_*, move_*_to_*, improve_*) is slightly inconsistent. Still, names are readable and predictable.
With 5 tools, the server is well-scoped for a library management and prompt improvement service. Each tool has a clear purpose and none are redundant.
The set covers listing, saving, organizing, and improving prompts. Missing delete/update operations for prompts or folders are minor gaps that agents can work around, but the core workflow is complete.
Your prompt library in any MCP client: list, fetch, save and version prompt templates.
Your versioned prompt library: search, read, save and version prompts from Claude Code.
Team prompt library with version control and reviews. Save, search and reuse prompts via MCP.
Read and write your Fresh Jots notes from Claude, Cursor, and any MCP client.
