Screenshots

The screenshot endpoint captures the session framebuffer as a PNG. Use it instead of page.screenshot().

Why a separate endpoint?

page.screenshot() reads pixels back through Chromium’s compositor. On remote sessions this path can stall or return empty frames. The /screenshot endpoint captures the framebuffer directly, avoiding the compositor entirely.

Taking a screenshot

$
curl -s https://api.chaser.sh/v1/sessions/{session_id}/screenshot \
>
  -X POST \
>
  -H "Authorization: Bearer $CHASER_KEY" \
>
  -o screenshot.png

The response is a raw PNG (image/png). No JSON wrapping.

From TypeScript:

1
const png = await fetch(
2
  `https://api.chaser.sh/v1/sessions/${sessionId}/screenshot`,
3
  {
4
    method: "POST",
5
    headers: { Authorization: `Bearer ${process.env.CHASER_KEY}` },
6
  }
7
).then((r) => r.arrayBuffer());
8
9
writeFileSync("screenshot.png", Buffer.from(png));

From Python:

1
import requests
2
3
png = requests.post(
4
    f"https://api.chaser.sh/v1/sessions/{session_id}/screenshot",
5
    headers={"Authorization": f"Bearer {os.environ['CHASER_KEY']}"},
6
).content
7
8
with open("screenshot.png", "wb") as f:
9
    f.write(png)

Response details

Timing

Screenshots take roughly 1–3 seconds. The session must be in ready status. If the session has expired, the endpoint returns 409.

Full-page screenshots

The endpoint captures the visible viewport. To capture the full page, scroll and stitch, or set the viewport size before navigating:

1
await page.setViewportSize({ width: 1280, height: 2000 });