api2ai

Language support and code generation for the .api2ai DSL.

The extension provides:

• Syntax highlighting
• Validation
• Auto completion
• Generate on save
• MCP host generation

api2ai generates curated MCP tools from existing OpenAPI specifications.

Requirements

• VS Code or Cursor 1.67+
• Node.js 20+ for generated projects and demo workspaces

Quick Start

1. Create a demo workspace

Open the Command Palette and run:

api2ai: Create demo workspace (MCP examples)

Choose an empty folder and open it when prompted.

2. Open the demo README

The generated workspace contains its own README.md with a guided walkthrough and several example projects.

The Open-Meteo example works out of the box and does not require an API key.

Working with your own APIs

Open a folder containing .api2ai files.

Whenever you save a file, the extension generates tool code and MCP hosts automatically.

Generated files are written to:

generated/api2ai/

Put API base URLs and secrets in the workspace .env (the env names from your DSL / host flags). Configure Cursor MCP in .cursor/mcp.json: stdio entries typically use envFile pointing at that .env; HTTP entries point at a running host URL (credentials stay with the host / .env, not in mcp.json).

Details: Cursor integration.

You can also trigger generation manually:

api2ai: Generate tool code (.ts + MCP host)

Documentation

Documentation index — architecture, authoring, runtime, and integrations.

Related Projects

• https://github.com/annettedorothea/core2ai
• https://github.com/annettedorothea/db2ai

Feedback

Share onboarding and integration feedback in GitHub Discussions. For bugs, open an Issue.

License

MIT — see LICENSE.

Whatever you do, work heartily, as for the Lord and not for men.

— Colossians 3:23

Created by Annette Pohl