OutEvalAI LogoOutEvalAI

Overview

Choose the right integration path - TypeScript SDK, Python SDK, or MCP server.

OutCallerAI provides typed SDKs for TypeScript and Python, plus a Model Context Protocol (MCP) server that lets AI coding agents interact with your workspace directly.

Integration paths

PathBest for
TypeScript SDKNode.js and Next.js backends, serverless functions, and full-stack apps
Python SDKPython services, data pipelines, ML workflows, and FastAPI backends
MCP serverAI coding agents that need to read workspace data and manage leads

What you can do

All three paths expose the same 23 public API operations across six resources:

ResourceOperations
WorkspaceGet workspace, update workspace
UsecasesList, get, create, update, delete usecases
AgentsList, get, create, update, delete voice agents
LeadsList, get, create, update, delete leads
CallsList, get, create calls
JobsCreate call export, get job status, cancel job

Common concepts

Authentication

All paths use the same API key. Create and manage keys from the Integrations page in your workspace dashboard.

Authorization: Bearer <api-key>

Base URL

https://api.outcallerai.com

Pagination

List endpoints support cursor-based pagination. Send limit (1-100), read X-Next-Cursor and X-Has-More from the response, then pass the cursor back on the next request. Stop when X-Has-More is false.

Idempotency

Create operations for agents and leads accept an optional Idempotency-Key header. When provided, the same key paired with the same request payload is safe to retry - the operation runs exactly once.

Retries

Both SDKs automatically retry transient failures (network errors, 5xx, timeouts) with exponential backoff. Retries are capped by elapsed time and attempt count. Only read-safe operations and idempotent writes are retried.

How is this guide?