# LAP - Lean API Platform

> A structured format that makes any API instantly usable by AI agents. One compiler, 1,000+ pre-compiled specs, and one-command skill generation.

LAP compiles API specifications (OpenAPI, Postman, AsyncAPI, GraphQL, Protobuf) into a structured, agent-native format. Browse 1,389 pre-compiled specs in the registry and generate plug-and-play agent skills.

## Key Features

- 88% fewer tokens compared to raw OpenAPI specs
- 35% cheaper API calls for AI agents
- 29% faster agent execution
- Verified across 500 benchmark runs with Claude Sonnet 4.5

## Three Pillars

### The Format
LAP is a structured, agent-native format that compiles any API spec into a compact representation. It preserves everything an agent needs to make the call while cutting token usage by up to 88%. Input formats: OpenAPI 3.x, Swagger 2.0, Postman Collections, AsyncAPI, GraphQL SDL, Protobuf, AWS Smithy.

### The Registry
A public registry of 1,389+ pre-compiled API specs ready for AI agents. Browse, search, and fetch LAP files at runtime. Free to use, free to redistribute.

- Registry: https://registry.lap.sh
- Machine-readable index: https://registry.lap.sh/llms.txt

### Skills
Curated bundles of related API endpoints grouped by task (e.g., "send a message", "create an invoice"). Skills wrap a LAP spec with auth setup, question routing, and execution playbooks. One command: `lapsh compile api.yaml --skill`.

## Getting Started

Install: `npm install -g @lap-platform/lapsh` or `pip install lapsh`

**API providers:** Run `lapsh compile` on your spec, then `lapsh publish` to the registry.
**Developers:** Browse the registry, download specs, or generate skills for Claude Code.
**Agent frameworks:** Fetch LAP files from the registry at runtime and feed them into your model's context.

## Links

- Website: https://lap.sh
- Registry: https://registry.lap.sh
- GitHub: https://github.com/Lap-Platform
- Full page content: https://lap.sh/llms-full.txt
- Deep technical context: https://lap.sh/llms-ctx.txt
- License: Apache 2.0