EXPLORE

Tools workbench

Queue a flow, then open a run to see, step by step, which URLs it hit, the raw responses it got and the facts it extracted.

API docs ↗

New run

Flow

Options

Save raw responses Keeps the body of every request so you can open it under Responses. Off: only the extracted facts are stored. Turn on when writing or debugging a projector. Always start a new run By default the same flow + params is deduplicated: an unfinished run is resumed, and one that succeeded inside the flow's TTL is reused (nothing runs). Tick to ignore both.

Capture network traffic browser steps only

While a browser step runs, JSON responses whose URL contains any of these substrings (one per line) are saved as <tool>.listen facts. * = everything except assets. Empty = off.

Kameleo profile optional

Browser fingerprint/session to open pages with. Leave empty to use the worker's default.

Runs execute in the tools worker.

Tool reference

Every step in a flow is one of these tools. emits = the fact keys it produces.

Run the worker from a terminal

The API only queues. The banner above can start the worker and Kameleo; otherwise:

Kameleo (Docker): scripts/kameleo-up.sh — needs KAMELEO_PAT in .env.

Runs

Live

when	flow	target	status	steps	facts

Select a run to see its steps, responses and facts.