Session replay

Recording a session

Recording is opt-in per session. Pass record: true when creating; nothing else about the session changes — same viewer, same CDP endpoint, same lifecycle.

curl -X POST https://sessions.browserview.io/sessions \  -H "authorization: Bearer $BROWSERVIEW_API_KEY" \  -H "content-type: application/json" \  -d '{"start_url": "https://example.com", "record": true}'

Capture happens outside the browser, so it works no matter who drives the session: a human in the live viewer, Playwright over CDP, or an agent. Recording adds no client-side scripts to the pages under test.

• Video — the X display encoded to WebM (VP8, 12 fps by default), including everything a user would have seen: dialogs, popups, downloads bars, cross-origin iframes.
• Actions — human input from the live viewer (mouse, keyboard, scroll) and automation commands sent over CDP (Input.*, Page.navigate, Runtime.evaluate), plus page navigations.
• Console — console.* calls and browser log entries from every tab.
• Network — one entry per request with method, URL, status, timing, and size. Sensitive headers (Authorization, Cookie, …) are redacted at capture time; bodies are never recorded.
• Errors — uncaught exceptions with stacks, and failed requests.

Fetching a replay

GET /sessions/{id}/replay returns {"status": "recording"} while the session is alive, 404 while the recording finalizes (typically under 30 seconds — poll every few seconds), and the full manifest once ready:

# Poll after the session ends (ready ~15-30s later)curl https://sessions.browserview.io/sessions/$SESSION_ID/replay \  -H "authorization: Bearer $BROWSERVIEW_API_KEY" {  "status": "ready",  "session_id": "1f0d33f9a1f2",  "started_at_ms": 1754038800000,  "ended_at_ms": 1754038923000,  "video": {    "url": "https://…",          # seekable WebM (VP8), ~1h presigned    "start_time_ms": 1754038800450,    "duration_ms": 122550,    "size_bytes": 15728640  },  "page_count": 2,  "pages": [    { "page_id": "p_000", "url": "https://example.com",      "start_time_ms": 1754038801000, "end_time_ms": 1754038860000 }  ],  "events": {    "actions":  { "url": "https://…", "count": 214, "size_bytes": 9182 },    "console":  { "url": "https://…", "count": 12,  "size_bytes": 2210 },    "network":  { "url": "https://…", "count": 96,  "size_bytes": 18876 },    "errors":   { "url": "https://…", "count": 1,   "size_bytes": 402 }  },  "urls_expire_at_ms": 1754042523000}

Artifact URLs are presigned and expire (see urls_expire_at_ms); re-fetch the manifest for fresh ones. The video is a single seekable WebM — a plain <video> element plays it, range requests make scrubbing cheap, and playbackRate gives you 2×/4× review.

Aligning events with video

// Every event line carries an absolute epoch-ms "ts".// One formula aligns everything with the video:videoTimeSeconds = (event.ts - manifest.video.start_time_ms) / 1000

# Event files are newline-delimited JSON (served gzip-encoded;# browsers and HTTP clients decompress transparently).{"ts":1754038801200,"src":"cdp","type":"page.navigated","url":"https://example.com"}{"ts":1754038802414,"src":"input","type":"input.btn","b":1,"d":true}{"ts":1754038802731,"src":"cdp","type":"network","method":"GET","url":"https://example.com/api","status":200,"durationMs":102,"encodedBytes":5120}{"ts":1754038803090,"src":"cdp","type":"console","level":"error","text":"Uncaught TypeError: x is not a function"}

The pages array is the main-frame navigation timeline across all tabs — useful for chaptering a replay or jumping to when a specific page was on screen.

Availability & retention

Replays also appear in the console: any recorded session gets a Replay entry under History with the full player — video, scrubber with per-page segments and error markers, and synchronized Actions / Console / Network / Errors panes.