LLMs & Agent Discovery

OpenSea provides machine-readable endpoints that help AI agents and LLMs discover and understand the API without reading human-formatted documentation.

llms.txt

The llms.txt file follows the llms.txt standard and provides a plain-text summary.

GET https://opensea.io/llms.txt

Use this as a starting point for any AI agent or LLM that needs to understand what the OpenSea API offers.

OpenAPI Specification

The OpenAPI specification provides a full machine-readable description of every endpoint, including request/response schemas, parameter types, and authentication requirements.

GET https://api.opensea.io/api/v2/openapi.json

This is the recommended entry point for agents that need to programmatically generate API calls, build client libraries, or validate request/response shapes.

When to use which

Token-optimized responses (TOON encoding)

Every GET endpoint on the OpenSea REST API supports TOON (Token-Optimized Object Notation), a compact, lossless text format that uses ~40% fewer tokens than JSON. Opt in by sending the Accept: text/markdown header:

# Standard JSON response (default)
curl "https://api.opensea.io/api/v2/collections/doodles-official" \
  -H "x-api-key: YOUR_API_KEY"

# TOON-encoded response (~40% fewer tokens)
curl "https://api.opensea.io/api/v2/collections/doodles-official" \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Accept: text/markdown"

TOON encodes JSON objects into a compact key-value text format with tabular arrays, nested indentation, and unquoted strings where safe. The response Content-Type will be text/markdown; charset=utf-8.

When to use it:

• AI agents processing API responses within limited context windows
• MCP servers or CLI tools where token efficiency matters
• Any client that can read structured plain text

JSON remains the default. Clients that don't send the Accept: text/markdown header are unaffected.

Instant API key

Agents can get a free-tier API key instantly with no signup:

POST https://api.opensea.io/api/v2/auth/keys

Returns a key with 60/min read and 5/min write rate limits, valid for 30 days. See Getting Your API Key for details.

Additional resources for agents