Skip to content
PrefactorPrefactorPrefactor Docs

Prefactor documentation

Prefactor records agent activity, classifies data risk, and stores quality evidence you attach, so you can review agents in production. Start with the quickstart, or go straight to the part of Prefactor you're working with.
QuickstartWhat is Prefactor

Get set up

Section titled “Get set up”
Run the quickstart

Register a demo agent, run it, and open its spans in the web app.

Set up with a coding agent

Install agent skills so Claude Code or Cursor can create Prefactor resources and instrument your agent.

Instrument your own agent

Add the TypeScript or Python SDK to an existing agent and pick the integration package for your framework.

Choose your tool

Section titled “Choose your tool”
TypeScript SDK

Integrations for LangChain, the Vercel AI SDK, and Claude.

Python SDK

Integrations for LangChain and LiveKit.

CLI

Manage accounts, agents, and judges from a terminal or CI.

API

No SDK needed. Call Prefactor over HTTP from any language, using the OpenAPI spec.

Work with your agents

Section titled “Work with your agents”
Monitor agents

Find an agent, open a run, and read what each span did.

Evaluate quality

Score runs with judges, or attach evaluations from your own pipeline.

Manage risk

Classify data risk with risk profiles and handle sensitive data.

Administer your account

Invite your team, issue API tokens, and set up environments.

Learn how Prefactor works

Section titled “Learn how Prefactor works”
What is Prefactor

How an agent moves from pilot to production on Prefactor.

Data model

How accounts, agents, instances, spans, and schemas fit together.

Concepts

One page for each Prefactor entity, from accounts to spans.