snaprender

SnapRender: see any web page

Real Chrome screenshots of any URL, saved as an image file you can look at and send to the user. Cookie consent pop-ups and ads are blocked by default. It also reads pages as clean markdown.

Setup (once)

1. Get a free key (200 captures a month, no card): https://snap-render.com/auth/signup
2. Put it in ~/.openclaw/openclaw.json:

   { skills: { entries: { snaprender: { apiKey: "sk_live_..." } } } }
   

   OpenClaw then sets $SNAPRENDER_API_KEY for every run. If your agent runs commands in a sandbox, also set SNAPRENDER_API_KEY in the sandbox environment (skill keys are not passed into sandboxes).

Capture a page

Use the exec tool with curl, not the browser tool: one request, no browser to drive, and the result is a ready image file.

Replace TARGET_URL and keep everything else as is:

mkdir -p snaprender && OUT="$PWD/snaprender/shot-$(date +%Y%m%d-%H%M%S)-$$.jpg"
code=$(jq -n --arg url 'TARGET_URL' \
  '{url: $url, format: "jpeg", quality: 80, block_ads: true, block_cookie_banners: true}' \
| curl -sS -X POST "https://app.snap-render.com/v1/screenshot" \
  -H "X-API-Key: $SNAPRENDER_API_KEY" -H "Content-Type: application/json" \
  -d @- -o "$OUT" -w '%{http_code}')
if [ "$code" = 200 ]; then echo "saved $OUT ($(wc -c < "$OUT" | tr -d " ") bytes)"; else echo "error $code: $(cat "$OUT")"; rm -f "$OUT"; fi

Then:

• Send it to the user with the message tool, giving the printed path as the media file. Never paste image data into the chat.
• Look at it yourself with view_image and the same path when the user asks what the page shows, whether it looks broken, or to compare pages.

Rules:

• Build the JSON with jq --arg as above. Never put user input straight into the shell string or the JSON.
• Use $SNAPRENDER_API_KEY exactly as written. Never print or echo the key.
• Each capture gets its own file (named by time), so comparisons never overwrite each other.

Options

Add fields to the jq object, for example {url: $url, device: "iphone_15_pro", dark_mode: true, ...}.