capture-screenshot

name	capture-screenshot
description	Capture a browser screenshot for task validation and attach it to the active task using attach_task_file.
allowed-tools	Bash(agent-browser:*), attach_task_file
metadata	{"author":"task-factory","version":"1.0","type":"follow-up","hooks":"pre-planning,pre,post"}

Capture Screenshot

Capture visual proof of the current app state and attach it to the task.

Workflow

1. Open the app

   • Prefer http://localhost:3000 unless task context specifies another URL.
   • If needed, start the app first and wait until it is reachable.

2. Take screenshot to a known path

mkdir -p .pi/tmp/screenshots
agent-browser open http://localhost:3000
agent-browser wait --load networkidle
agent-browser screenshot .pi/tmp/screenshots/validation-$(date +%s).png

3. Attach screenshot to task
   • Use attach_task_file with the screenshot path.
   • Include taskId if you can read it from the task context; otherwise omit it.

Example tool call:

attach_task_file({
  path: ".pi/tmp/screenshots/validation-12345.png",
  taskId: "<TASK_ID>",
  filename: "validation-screenshot.png"
})

4. Report result
   • Confirm which URL was captured.
   • Confirm the attachment filename/ID returned by attach_task_file.

Notes

• Use a deterministic screenshot path (don’t rely on unknown temp output paths).
• If capture or attach fails, explain exactly what failed and stop.
• You may take multiple screenshots if needed, but attach at least one clear validation image.