Public API Skills

A collection of public API documentation converted to Agent Skills format - structured markdown optimized for AI agents.

Why Agent Skills?

AI agents often need to interact with external APIs. Raw OpenAPI specifications have limitations:

• Too large - Complex specs can be megabytes, exceeding LLM context limits
• Wasteful - Loading the entire document for every request wastes context
• Monolithic - All-or-nothing access prevents selective loading

Agent Skills solves this by splitting documentation into on-demand files. Agents load only what they need - start with an overview, drill into specific operations.

No special integrations required. File reading works in every agent framework.

Available APIs

Usage

Point your AI agent to read apis///SKILL.md as the entry point.

Example for Stripe:

apis/stripe/stripe-api/SKILL.md

How It Works

Each API is converted using openapi-to-skills:

npx openapi-to-skills ./openapi.json -o ./output

The tool transforms monolithic OpenAPI specs into a navigable file structure:

/
├── SKILL.md                    # Entry point with overview
└── references/
    ├── resources/              # One file per resource/tag
    ├── operations/             # One file per operation
    ├── schemas/                # Grouped schema files
    └── authentication.md

Customizing Generated Skills

Generated files are overwritten on each generate:* run. To add API-specific content that survives regeneration, use custom templates and extras:

apis//
├── openapi.json          # Source spec
├── templates/            # Custom Eta templates (partial override)
│   ├── skill.md.eta      # Only override templates you need to change
│   └── operation.md.eta  # Falls back to defaults for the rest
├── extras/               # Static files copied into output post-generation
│   └── references/