replayio

name	replayio
description	Use when you need to record or inspect a Playwright CLI browser run in Replay, test a local app directly with Playwright CLI, or use the Replay MCP server for deeper debugging of an uploaded recording.

Replay Browser + Playwright CLI + Replay MCP

Use playwright-cli with the Replay Browser whenever you need a recorded browser session. Recordings upload automatically — you do not need to run replayio upload yourself. Your only responsibility is to close the browser when the test is complete. The plugin's hooks handle the rest:

• Close hook: when you run a playwright-cli close, pending recordings upload immediately.
• Stop/idle hook: when your agent turn ends or idles, any still-open browser is closed and pending recordings upload as a safety net.

Always prefer closing explicitly — it gives faster feedback and ensures recordings finish writing before you report results.

Direct CLI first

Use playwright-cli directly for live browser control and first-pass inspection. Do not reach for the Replay MCP server just to click, type, snapshot, read console output, inspect network requests, take screenshots, read storage, or check cookies.

Use the Replay MCP server only after a recording has uploaded and you need deeper Replay-specific debugging, such as inspecting execution history, narrowing a time-travel debugging problem, or investigating details that the live CLI cannot answer.

Useful direct commands:

"$PWCLI" --session="$SESSION" snapshot
"$PWCLI" --session="$SESSION" console
"$PWCLI" --session="$SESSION" console error
"$PWCLI" --session="$SESSION" network
"$PWCLI" --session="$SESSION" screenshot output/playwright/$SESSION/page.png
"$PWCLI" --session="$SESSION" eval "document.title"
"$PWCLI" --session="$SESSION" localstorage-list
"$PWCLI" --session="$SESSION" sessionstorage-list
"$PWCLI" --session="$SESSION" cookie-list

Close-when-done contract

After you finish a test run, before reporting the outcome to the user, close the browser:

"$PWCLI" --session="$SESSION" close

Do not leave a browser open at the end of your turn. If you forget, the stop/idle hook will clean up, but you may not see the resulting Replay URL before responding.

The reliable path

1. Verify npx exists.
2. Verify the Replay runtime exists and the CLI is logged in.
3. Verify .playwright/cli.config.json points at the Replay Chromium binary.
4. Set both RECORD_ALL_CONTENT='1' and RECORD_REPLAY_VERBOSE='1'.
5. If testing a local app, start it first and verify the actual reachable URL.
6. Use --session=<short-name>, not -s=....
7. Keep session names short on macOS. Long names can fail with listen EINVAL ... .sock.
8. Drive and inspect the page directly with playwright-cli commands.
9. Close the browser with "$PWCLI" --session="$SESSION" close when done — uploads happen automatically.

Prerequisites

Check npx first:

command -v npx >/dev/null 2>&1

If missing, stop and ask the user to install Node.js/npm.

Replay MCP authentication (private recordings)

Replay MCP calls are authorized per user. If tools return Access denied, you are usually not authenticated to Replay as the same account that owns the recording.

• Cursor: open Settings → MCP, select the Replay server, and complete Sign in / Connect / Reconnect so Cursor can finish the OAuth flow (including dynamic client registration when supported). Stay on https://dispatch.replay.io/mcp (do not swap in unrelated endpoints unless Replay explicitly instructs you to).
• Other hosts: follow that host’s MCP authentication mechanism (some environments use API keys or separate app connectors instead of OAuth).

Replay setup

Do not blindly reinstall Replay on every run. Verify first.

replayio info
replayio whoami

If Replay is missing, install it:

npx @replayio/replay install

If not logged in, authenticate:

replayio login

Playwright CLI config

playwright-cli should use the Replay Browser through .playwright/cli.config.json:

{
  "browser": {
    "launchOptions": {
      "executablePath": "/Users/YOURUSERNAME/.replay/runtimes/Replay-Chromium.app/Contents/MacOS/Chromium"
    }
  }
}

Do not rely on a global --executable-path flag — the config file is the reliable way to select Replay Chromium.

Skill path

The plugin ships a wrapper script so you can invoke playwright-cli without a global install:

export RECORD_ALL_CONTENT='1'
export RECORD_REPLAY_VERBOSE='1'
export PWCLI="${CLAUDE_PLUGIN_ROOT:-$PWD/dist/claude-code/replayio}/skills/replayio/scripts/playwright_cli.sh"

If the plugin root environment variable is not set, run from the generated plugin repo root or point PWCLI at the Replay plugin checkout.

Local app check

If testing a local app:

1. Start the app first.
2. Use the URL the dev server actually prints.
3. Do not assume the requested port is the final port. Some dev servers auto-increment when the port is busy.
4. Verify reachability before opening the browser.

curl -I http://127.0.0.1:4323/todos

If a localhost request fails even though a process is clearly listening, you may be in a restricted sandbox. Rerun the browser and reachability checks outside the sandbox.

Reliable session workflow

SESSION=rt
URL=http://127.0.0.1:4323/todos
mkdir -p "output/playwright/$SESSION"

"$PWCLI" --session="$SESSION" open "$URL" --headed --browser=chrome
# wait for: Browser `rt` opened with pid ...

"$PWCLI" --session="$SESSION" snapshot
"$PWCLI" --session="$SESSION" fill e6 "Buy milk"
"$PWCLI" --session="$SESSION" click e7
"$PWCLI" --session="$SESSION" snapshot
"$PWCLI" --session="$SESSION" console error
"$PWCLI" --session="$SESSION" screenshot output/playwright/$SESSION/after-add.png

# When the test is complete, close the browser.
# The upload hook will upload the recording automatically.
"$PWCLI" --session="$SESSION" close

Use snapshot before using element refs, and again after DOM changes or navigation.

Attach to an already-open browser

If the browser is already running:

"$PWCLI" list
"$PWCLI" attach rt --session=rt1
"$PWCLI" --session=rt1 snapshot

Analyzing uploaded recordings

First inspect the live run with direct playwright-cli commands. Once a recording has uploaded, use the replay MCP server tools only when you need deeper Replay-specific debugging beyond direct CLI output. Claude Code connects to the Replay HTTP MCP server configured in .mcp.json.

Recording uploads are automatic

You do not need to run replayio list or replayio upload <id> manually. The plugin's hooks handle uploads:

• When you close a session ("$PWCLI" --session=... close), the close hook uploads any new recordings.
• When your turn ends or idles with a browser still open, the stop/idle hook closes the session and uploads.

If you need to inspect the upload state yourself:

replayio list

Troubleshooting

• If open appears stuck, wait for the browser-opened output before sending the next command.
• If snapshot says the browser is not open, retry with --session=<name> instead of -s=....
• If you see listen EINVAL ... .sock, shorten the session name.
• If the app is on localhost, verify the exact URL with curl -I before opening the browser.
• If the requested port was busy, use the actual port printed by the dev server.
• If a browser exists but your commands are detached, use list then attach.
• Prefer explicit commands like fill, click, check, and press over eval or run-code.
• Prefer direct CLI inspection (console, network, screenshot, storage, cookies, eval) before using the Replay MCP server.
• If Replay authentication fails, run replayio login or reconnect the relevant Replay app/integration.

References

• CLI command reference: references/cli.md
• Workflow notes: references/workflows.md
• Replay docs