Generate curated MCP tools from OpenAPI specifications.
| Repository | Purpose |
|---|---|
| core2ai | Shared runtime, architecture, and documentation |
| api2ai | Generate curated MCP tools from OpenAPI specifications |
| db2ai | Generate curated MCP tools from relational databases |
Instead of hand-writing MCP servers, pick the OpenAPI operations you want as tools, enrich them in .api2ai, and generate executable MCP tooling.
You curate which endpoints become tools — not every path in the specification is exposed automatically.
The easiest way to explore api2ai is with the VSIX extension and the bundled demo workspace.
Download the latest VSIX from the releases page:
https://github.com/annettedorothea/api2ai/releases
In Cursor or VS Code, run:
api2ai: Create demo workspace (MCP examples)
Open the generated Demo Workspace README and follow the Quick Start.
The Open-Meteo example works out of the box and does not require an API key.
OpenAPI Specification
│
▼
select & enrich
│
▼
.api2ai
│
▼
Generated MCP Server
│
▼
Cursor • ChatGPT • Claude • MCP Inspector
Example:
openapi "./openapi/github-user-min.openapi.yaml"
auth {
in: header
name: "Authorization"
prefix: "Bearer "
}
GET "/user" {
toolName: getGitHubAuthenticatedUser
access: protected
summary: "Get the authenticated user"
}
Documentation index — architecture, authoring, runtime, and integrations.
See CHANGELOG.md for version history and upgrade notes.
- core2ai — Shared runtime, code generation infrastructure, and documentation.
- db2ai — Generate curated MCP tools from relational databases.
We welcome feedback on onboarding, documentation, DSL ergonomics, and MCP integration. Share your experience in GitHub Discussions. For bugs, open an Issue.
MIT — see LICENSE.
Whatever you do, work heartily, as for the Lord and not for men.
— Colossians 3:23
Created by Annette Pohl