Tool activity

Response snapshots include activity from their execution turns. Use fal.activities to display progress and fal.activated_skills to inspect loaded skills.

import { agent } from "./client.ts";

for await (const response of agent.stream({
  input: "Plan a product photo.",
})) {
  for (const activity of response.fal.activities ??
    []) {
    console.log(
      activity.id,
      activity.label,
      activity.status,
    );
  }
  console.log(response.fal.activated_skills ?? []);
}

Each update contains the complete activity snapshot. Replace displayed state instead of appending every update. Activity IDs remain stable when you retrieve or reconnect to the same response.

Field	Meaning
`id`	Stable activity identity within the response.
`turn_id`	Execution turn that produced the activity.
`kind`	`tool` or `reasoning`.
`label`	Display label.
`status`	`running` or `completed`.
`started_at`, `ended_at`	Unix timestamps in milliseconds. The end time is optional.
`duration_ms`	Optional duration in milliseconds.

Tool activities include node, an optional tool name, and optional JSON arguments and output. Reasoning activities include visible reasoning in content. Check the activity kind before reading its fields. Use structured input requests for questions and approvals. Read the response status and error to determine whether the task succeeded. activated_skills contains names loaded during these turns, including active skills inherited from the conversation. It does not list every available skill. See skills for discovery and activation. Both arrays are optional. Missing fields mean the response does not report tool or skill activity.

Setting Up

Model APIs

Serverless

Compute

Organizations

fal Agent

fal Assets

Tool activity