lapsh 0.4.8

Lean API Platform -- Token-efficient API specs for AI agents

Lean API Platform

Agent-Native API specs. Verified, compressed, ready to install.

Website · Registry · Benchmarks · Docs

---

Without API documentation, LLM agents hallucinate endpoints, invent parameters, and guess auth flows -- scoring just 0.399 accuracy in blind tests.

LAP fixes this. One command gives your agent a verified, agent-native API spec -- jumping accuracy to 0.860. And because LAP specs are up to 10x smaller than raw OpenAPI, you also save 35% on cost and run 29% faster.

Not minification -- a purpose-built compiler with its own grammar.

Proven in 500 blind runs across 50 APIs

LAP Lean scored 0.851 (vs 0.825 raw) while using 35% less cost and 29% less time -- same accuracy, far fewer tokens.

Full benchmark report (500 runs, 50 specs, 5 formats) · Benchmark methodology and data

Quick Start

# Search the registry for an API
npx @lap-platform/lapsh search payment

# Download a spec
npx @lap-platform/lapsh get stripe-com -o stripe.lap

# Install a pre-compiled skill (drops into ~/.claude/skills/)
npx @lap-platform/lapsh skill-install stripe-com

# Or compile your own spec
npx @lap-platform/lapsh compile api.yaml --lean

Use as an Agent Skill

Install the LAP skill so your agent can search, compile, and manage APIs automatically:

Claude Code:

cp -r skills/lap ~/.claude/skills/lap

OpenClaw: install from ClawHub or copy manually:

cp -r skills/lap ~/.openclaw/skills/lap

Once installed, agents auto-trigger the skill when working with APIs -- or invoke it directly with /lap. You can also install individual API skills for specific integrations:

npx @lap-platform/lapsh skill-install stripe-com
# Agent now knows the full Stripe API

Want to get listed? Register as a verified publisher and share your specs and skills with the registry.

# Install globally (npm or pip)
npm install -g @lap-platform/lapsh
pip install lapsh

What You Get

• 📦 Registry — browse and install 1500+ pre-compiled specs at lap.sh
• 🗜️ 5.2× median compression on OpenAPI, up to 39.6× on large specs — 35% cheaper, 29% faster (benchmarks)
• 📐 Typed contracts — enum(a|b|c), str(uuid), int=10 prevent agent hallucination
• 🔌 6 input formats — OpenAPI, GraphQL, AsyncAPI, Protobuf, Postman, Smithy
• 🎯 Zero information loss — every endpoint, param, and type constraint preserved
• 🔁 Round-trip — convert back to OpenAPI with lapsh convert
• 🤖 Skill generation — lapsh skill creates agent-ready skills from any spec
• 🔗 Integrations — LangChain, Context Hub, Python/TypeScript SDKs

How It Works

Five compression stages, each targeting a different source of token waste:

Stage	What it does	Savings
Structural removal	Strip YAML scaffolding — paths:, requestBody:, schema: wrappers vanish	~30%
Directive grammar	@directives replace nested structures with flat, single-line declarations	~25%
Type compression	type: string, format: uuid → str(uuid)	~10%
Redundancy elimination	Shared fields extracted once via @common_fields and @type	~20%
Lean mode	Strip descriptions — LLMs infer meaning from well-named parameters	~15%

Benchmarks

1,500+ specs · 5,228 endpoints · 4.37M → 423K tokens

Format	Specs	Median	Best
OpenAPI	30	5.2×	39.6×
Postman	36	4.1×	24.9×
Protobuf	35	1.5×	60.1×
AsyncAPI	31	1.4×	39.1×
GraphQL	30	1.3×	40.9×

Verbose formats compress most — they carry the most structural overhead. Already-concise formats like GraphQL still benefit from type deduplication.

The Ecosystem

LAP is more than a compiler:

Component	What	Command
Search	Find APIs in the registry	lapsh search payment
Get	Download a spec by name	lapsh get stripe-com
Skill Install	Install a pre-compiled skill	lapsh skill-install stripe-com
Compiler	Any spec → .lap	lapsh compile api.yaml
Skill Generator	Create agent-ready skills from any spec	lapsh skill api.yaml --install
API Differ	Detect breaking API changes	lapsh diff old.lap new.lap
Round-trip	Convert LAP back to OpenAPI	lapsh convert api.lap -f openapi
Publish	Share specs to the registry	lapsh publish api.yaml --provider acme

Claude Code users: The lap skill is included -- your agent can search and install APIs directly. See skills/lap/SKILL.md.

Supported Formats

lapsh compile  api.yaml           # OpenAPI 3.x / Swagger
lapsh compile  schema.graphql     # GraphQL SDL
lapsh compile  events.yaml        # AsyncAPI
lapsh compile  service.proto      # Protobuf / gRPC
lapsh compile  collection.json    # Postman v2.1
lapsh compile  model.smithy       # AWS Smithy

Format is auto-detected. Override with -f openapi|graphql|asyncapi|protobuf|postman|smithy.

Top Compressions

Integrations

# LangChain
from lap.middleware import LAPDocLoader
docs = LAPDocLoader("stripe.lap").load()

LangChain, Context Hub, and Python/TypeScript SDKs. See integration docs.

FAQ

Why do agents hallucinate API calls?

Because they have no way to find the spec, and even if they could, it's a million tokens of YAML written for humans. Agents without specs score 0.399 accuracy -- wrong 60% of the time. They hallucinate endpoint paths, send invalid types, and miss auth. Give them a LAP spec and accuracy jumps to 0.860. The spec doesn't make the agent smarter. It makes guessing unnecessary.

How is this different from OpenAPI?

LAP doesn't replace OpenAPI — it compiles FROM it. Like TypeScript → JavaScript: you keep your OpenAPI specs, your existing tooling, everything. LAP adds a compilation step for the LLM runtime.

How is this different from MCP?

MCP defines how agents discover and invoke tools (the plumbing). LAP compresses the documentation those tools expose (the payload). They're complementary — LAP can compress MCP tool schemas.

Why not just minify the JSON?

Minification removes whitespace — that's ~10% savings. LAP performs semantic compression: flattening nested structures, deduplicating schemas, compressing type declarations, and stripping structural overhead. That's 5-40× savings. Different class of tool.

What about prompt caching?

Use both. Compress with LAP first, then cache the compressed version. LAP reduces the first-call cost and frees context window space. Caching reduces repeated-call cost. They stack.

Will LLMs understand this format?

Yes. LAP uses conventions LLMs already know — @directive syntax, {name: type} notation, HTTP methods and paths. In blind tests, agents produce identical correct output from LAP and raw OpenAPI. The typed contracts actually reduce hallucination.

What if token costs keep dropping?

Cost is the least important argument. The core value is typed contracts: enum(succeeded|pending|failed) prevents hallucinated values regardless of token price. Plus: formal grammar (parseable by code, not just LLMs), schema diffing, and faster inference from fewer input tokens.

Contributing

See CONTRIBUTING.md. The test suite has 1,101 tests across 190+ example specs in 6 formats.

git clone https://github.com/Lap-Platform/lap.git
cd lap
pip install -e ".[dev]"
pytest

License

Apache 2.0 — See NOTICE for attribution.

---

lap.sh · Built by the LAP team