Skip to main content

​
What is MCP?

The Model Context Protocol (MCP) is a standardized JSON-RPC 2.0 protocol that enables AI agents to communicate with external data sources and tools. PromptRepo implements an MCP server that exposes your prompt library directly to AI assistants like Claude Code and Claude Desktop.
Think of MCP as a real-time API for AI agents — instead of copy-pasting prompts manually, your AI assistant can query, search, and resolve prompts on-demand during conversations.

​
Why Use PromptRepo’s MCP Server?

Zero-Friction Access

AI agents can fetch prompts without leaving their environment — no browser tab switching or copy-paste required.

Dynamic Resolution

Variables like {{language}} or {{framework}} are resolved at runtime, making prompts reusable across contexts.

Full-Text Search

Agents can search your entire prompt library using natural language queries.

Version Control

Always get the latest version of a prompt — updates propagate instantly to all connected agents.

​
How It Works

1

Agent sends JSON-RPC request

The AI assistant calls the MCP endpoint at POST /api/mcp with a tool name and parameters.
2

Authentication via API key

PromptRepo verifies the API key from the Authorization or x-api-key header.
3

Tool execution

The dispatcher routes the request to the appropriate handler (list, get, resolve, or search).
4

Response returned

Results are sent back to the agent as a JSON-RPC 2.0 response envelope.

​
Available Tools

PromptRepo’s MCP server exposes four core tools:
Returns paginated prompt metadata (title, description, variables) without full content. Useful for browsing your library.Parameters:
  • limit (optional): Max results (1-100, default 20)
  • offset (optional): Pagination offset (default 0)
Access:
  • Authenticated: Your prompts + public prompts
  • Anonymous: Public prompts only
Fetches a single prompt by ID, including full content and extracted variable names.Parameters:
  • prompt_id (required): UUID of the prompt
Access:
  • Authenticated: Your own prompts (public/private) + others’ public prompts
  • Anonymous: Public prompts only
Fetches a prompt and substitutes {{variable}} placeholders with provided values. Unmatched variables remain as-is.Parameters:
  • prompt_id (required): UUID of the prompt
  • variables (optional): Map of variable names to replacement values
Returns:
  • resolved_content: Template after substitution
  • unresolved_variables: Variables without provided values
Full-text search over accessible prompts using PostgreSQL’s tsvector indexing.Parameters:
  • query (required): Search string (min 1 character)
  • limit (optional): Max results (1-50, default 10)
Access:
  • Authenticated: Search your prompts + public prompts
  • Anonymous: Search public prompts only

​
Authentication Model

PromptRepo uses API key authentication for MCP requests:
  • Authenticated users: Generate an API key from /profile. Access your private prompts + all public prompts.
  • Anonymous users: No API key required. Access public prompts only (read-only).
API keys are bearer tokens — treat them like passwords. Never commit them to version control or share them publicly.

​
Next Steps

Authentication Setup

Generate and manage API keys for secure access

Configuration

Connect Claude Desktop, Claude Code, and other MCP clients

​
Technical Specification

  • Protocol: JSON-RPC 2.0 over HTTP
  • Endpoint: POST /api/mcp
  • Content-Type: application/json
  • CORS: Enabled for cross-origin requests
  • Auth headers: Authorization: Bearer <key> or x-api-key: <key>
  • Response format: Always HTTP 200 (errors in JSON-RPC envelope)
The MCP endpoint is excluded from session-based authentication middleware, so you must use API key headers — browser cookies are not supported.

Build docs developers (and LLMs) love

Get started for freeTalk to us