Skip to navigation
Getting Started

Introduction

Runtype is an AI product development platform. Use the Runtype API to manage Agents, Flows, Tools, Records, Products, Surfaces, Evals, Schedules, API keys, Model Configs, and Secrets. This section covers programmatic access through the REST API, the SDKs, and the runtype CLI. For dashboard workflows and a product overview, see What is Runtype? in the User Guide.

API capabilities

Use the API to do the following:

  • Build Agents and Flows: Create tool-using Agents and multi-step Flows with prompts, context steps, and conditions.
  • Run Agents and Flows: Send a saved or inline definition to POST /v1/dispatch and stream the results as server-sent events (SSE), or process Records in batches.
  • Deploy through Surfaces: Expose Agents and Flows through chat widgets, REST APIs, Model Context Protocol (MCP) servers, Agent-to-Agent (A2A) endpoints, and messaging channels.
  • Store Records: Keep structured data that Agents and Flows use.
  • Test with Evals: Run Agents and Flows against a suite of test cases and score the output.
  • Use Tools: Create custom Tools or use built-in Tools.
  • Use supported models: Run Agents and Flows with models from supported providers, including OpenAI, Anthropic, Google, and xAI.

Use these links to start with authentication, your first request, the API reference, or an SDK.

Authentication

Create an API key, or let an agent self-register and claim a key

Quickstart

Define and run your first flow

API Reference

Find endpoints and request details

Client libraries

Install the TypeScript, Python, or Ruby SDK

Base URL

Send API requests to the following base URL:

Base URL
https://api.runtype.com/v1

The runtype CLI and the official SDKs connect to api.runtype.com over HTTPS. If you run them in CI, a sandbox, or another restricted-egress environment, allowlist that host. A proxy or firewall that intercepts the connection can report an authentication failure instead of a network error. Check reachability before you treat the API key as invalid.

Response format

Runtype resource responses use camelCase field names. The following response shows a single Flow:

Response
{
  "id": "flow_01k9x2q0000000000000000000",
  "name": "Customer Analysis",
  "createdAt": "2024-01-15T10:30:00Z",
  "updatedAt": "2024-01-15T10:30:00Z",
  "lastRunAt": null,
  "lastModifiedSource": "dashboard"
}

List endpoints return results in data with a pagination object. To get the next page, pass pagination.nextCursor as the cursor query parameter. Errors use standard HTTP status codes, and most return an error message. Validation errors add a details array. For more information, see Common patterns.

Execution requests return 429 with a Retry-After header when you exceed your plan’s burst rate, and 402 when a spending control applies. For more information, see Rate limits.

Get help

If you need help with the API, use one of these channels:

Next steps

Continue with one of these developer guides:

Working with tools

Configure and call Tools

Runtime tools

Define Tools inside dispatch requests

Bring your own agent

Connect an agent that runs outside Runtype

End-user identity and credentials

Recognize end users and connect their third-party accounts

Runtype MCP server

Manage Runtype from an MCP client

Connect Runtype to coding agents

Set up Claude Code, Cursor, Codex, and other coding agents with the CLI

Run flows and agents with dispatch

Run a Flow or Agent with one API call