For AI Agents

How AI agents can interact with the Arke API—authentication, entity operations, and LLM-friendly reference formats.

Overview

Arke is designed to be agent-friendly. AI agents can create, read, update, and search entities on the network through the REST API. The system includes specific features for agent interaction:

API key authentication for agent services
Condensed operations reference optimized for LLM context windows
Structured entity model that maps cleanly to function calling
Temporal permissions for scoped, time-limited access

Getting Started as an Agent

1. Get an API Key

Agents authenticate with API keys:

Authorization: ApiKey uk_xxxxx

2. Fetch the Operations Reference

The condensed API reference is available at:

GET https://api.arke.institute/ops-reference

This returns a plain-text reference of all API operations, optimized for LLM consumption (~80% fewer tokens than the full OpenAPI spec). It includes:

All endpoints with methods and paths
Required and optional parameters
Request body fields
Permission requirements
Authentication modes

3. Get the Full OpenAPI Spec

For structured integration, the complete OpenAPI 3.1 specification is available at:

GET https://api.arke.institute/openapi.json

LLM-Friendly Resources

Resource	URL	Format
Operations Reference	`https://api.arke.institute/ops-reference`	Plain text (condensed)
OpenAPI Spec	`https://api.arke.institute/openapi.json`	JSON (complete)
Documentation Index	`https://docs.arke.institute/llms.txt`	Markdown (structured links)
Full Documentation	`https://docs.arke.institute/llms-full.txt`	Markdown (concatenated)