Documentation
Agent Flow Bind docs
Guides and an API reference for integrating with Agent Flow Bind — creating agents, running them, handling human-in-the-loop steps, and verifying webhooks.
Getting started
Concepts
Runs lifecycle & statuses
The run status machine, endpoints for each stage, and the exact JSON shapes.
Presets & schemas
The preset catalogue, what GET /v1/presets returns, bringing your own schema, and agent versioning.
Human-in-the-loop
Answering, approving and reviewing a paused run: the three paused statuses, their resume endpoints and the answers they expect.
Knowledge
Uploading reference documents (txt, csv, md, pdf, up to 25 MB) that an agent can search while it runs.
Evals
Turning corrected output into regression tests for an agent, and running them against its current or a past version.
Channels
Data sources
Connecting a database or a file store read-only: tables an agent consults during a run, documents read in, results delivered back.
Connections
Delivery targets a run's output can be saved to, beyond the default download.
Webhooks
Subscribing to run events, the events a webhook can fire on, and verifying each delivery's signature.
Published apps
Letting an end user run one agent through a branded, no-API-key web app.
Reference
SDK (TypeScript)
The @agentflowbind/sdk TypeScript package, in preview: the intended API surface, so you can plan an integration before it ships.
Errors & rate limits
The error envelope, common error codes, idempotency and rate limits of the Agent Flow Bind API.
Data & providers
Which model providers process your data, where, and how it's disclosed.
API reference
Every endpoint, grouped by resource, generated from the API's own OpenAPI document.