MCP Overview
What MCP is, why Noukai uses it, and how it enables AI-assisted flow building.
What is MCP?
The Model Context Protocol (MCP) is a standard for AI assistants to interact with external tools. It defines how tools are discovered, called, and how results are returned.
Why Noukai Uses MCP
Noukai exposes 61 MCP tools that let AI assistants build and manage flows programmatically. Instead of a user clicking through a UI, they describe what they want and the AI handles it.
Architecture
- Transport: Streamable HTTP (stateless, per-request auth)
- Auth: OAuth 2.1 + PKCE (browser sign-in, auto-refresh)
- Server URL:
https://api.noukai.dev/api/v1/mcp
Tool Categories
| Category | Count | Purpose |
|---|---|---|
| Read | 11 | Inspect flows, blocks, evals, history, run traces |
| Structural | 11 | Add, remove, move, wrap blocks |
| Lifecycle | 2 | Create and delete flows |
| Content | 9 | Update prompts, schemas, configs (incl. atomic multi-field) |
| Block Checkpoint | 2 | Save and delete block-prompt snapshots |
| Eval | 9 | Author judges, attach them as bindings, snapshot judge bodies |
| Versioning | 2 | Publish and set production |
| Object Model | 3 | Reusable schema types |
| Test Case | 9 | Create, manage, and execute tests |
| Authoring Guide | 3 | Fetch the pipeline pattern/recipe corpus |
Next Steps
- Setup — configure your MCP client
- Tool Listing — all 61 tools with schemas
- Common Workflows — step-by-step recipes