AI Agent Integration Guide

One-page API reference: API for AI Agents (auth, discovery, read, limits, errors).

PipeAgent is built so agents consume production-ready JSON, not raw HTML: creators monetize insights as APIs; your workflows stay on stable schemas and metered endpoints. This guide wires that into OpenClaw, LangChain, AutoGPT-style stacks, and more.

1. Agent-first design

Unlike ad hoc REST for dashboards, PipeAgent prioritizes:

2. Authentication

Your agent should manage API keys securely. We recommend using Environment Variables.

const response = await fetch('https://api.pipeagent.dev/v1/feed/your-feed-id', {
  headers: {
    'X-API-KEY': process.env.PIPEAGENT_API_KEY
  }
});

3. Recommended Workflow for Agents

Step A: Discovery

If the agent doesn't know where to find specific data, query search or catalog with your read key (x-api-key or Authorization: Bearer):

Step B: Metadata (no data payload)

Call GET https://api.pipeagent.dev/v1/feed/{id}/metadata (same read key; {id} is the feed UUID or endpoint_path) to read consumer_type, schema_definition, sample_response, pricing hints, and provider summary—without executing a full /v1/feed/{id} data fetch.

Step C: Consumption

Fetch the data with GET /v1/feed/{id}. PipeAgent applies schema validation and gateway semantics so the payload your agent sees stays consistent with metadata and docs.

4. Framework Integration Guides

PipeAgent supports all major AI Agent frameworks. Choose your framework below for a detailed tutorial:

5. Best Practices