Documentation
API Reference
Promptmark offers three integration surfaces, each suited to different use cases.
Integration Tiers
| Tier | Surface | Auth | Audience |
|---|---|---|---|
| 1. MCP Tools | 74 tools via stdio or HTTP | OAuth 2.0 device flow / Bearer JWT | AI agent developers |
| 2. REST API | /api/* JSON endpoints |
Bearer JWT (Authorization: Bearer <token>) |
App integrators |
| 3. OAuth Discovery | RFC 9728 / 8414 endpoints | N/A (public metadata) | MCP client developers |
Which Should I Use?
- Building an AI agent? Use MCP Tools. They’re the primary integration point and cover all prompt management operations.
- Building an app integration? Use the REST API. Standard HTTP with JSON request/response bodies.
- Building an MCP client? Use the OAuth Integration guide for discovery and authentication protocols.
Authentication
All API access requires a Bearer JWT token in the Authorization header:
Authorization: Bearer eyJhbGciOiJIUzI1NiIs...See Authentication for how to obtain tokens via OAuth, device flow, or cookie session.
Guides
- Authentication — JWT tokens, OAuth, device flow
- MCP Setup — Connect AI clients to Promptmark’s MCP server
- MCP Reference — All 74 MCP tools documented
- REST Reference — REST API endpoints
- OAuth Integration — RFC 9728/8414, PKCE, dynamic client registration
Authentication
How to authenticate with Promptmark's API using JWT tokens, OAuth, and device flow.
MCP Setup
Connect AI clients and agent frameworks to Promptmark's MCP server — Claude.ai, ChatGPT, Claude Code, Cursor, VS Code, Warp, OpenCode, CrewAI, LangChain, Pydantic AI, and more.
MCP Reference
Complete reference for the Promptmark MCP tool catalog.
REST Reference
REST API endpoints for Promptmark.
OAuth Integration
RFC 9728/8414 OAuth integration for MCP clients connecting to Promptmark.