Agent task

Research the web

Find, inspect, extract, and synthesize reliable information from the open web.

Target outcome

A source-backed research result with clean evidence and durable context.

Required capabilities

Choose one tool for each capability, then connect them as a workflow.

Find relevant, current sources with structured results that an agent can rank and cite.

Navigate interactive sites, execute actions, and collect evidence in a managed browser.

Turn pages and documents into clean, structured, model-ready content.

Store, retrieve, and rank durable context for agents across sessions.

Trace agent runs, evaluate outcomes, and diagnose failures in production.

A source-backed research pipeline that searches, browses, extracts, synthesizes, and remembers.