debugging-react-native

<objective> Connect to running React Native apps via Metro bundler to read console logs and execute JavaScript. Provides three tools for app discovery, log streaming, and JS execution in the app context. </objective>

<quick_start> Prerequisite: pip install websocket-client Assume a single app is connected. Just call read_logs.py or execute_in_app.py directly - they auto-select the app when only one is connected (the typical case).

# Read recent logs
python scripts/read_logs.py

# Execute JS in app
python scripts/execute_in_app.py "Date.now()"

Only use discover_apps.py or --app-id when you get an error about multiple apps being connected. </quick_start>

<tools> <tool name="read_logs.py"> Stream console logs from app.

python scripts/read_logs.py [--max-logs 100] [--filter "error|warn"] [--timeout 5]

• --filter: Regex to filter log messages
• --timeout: Seconds to collect logs (default: 5)
• --app-id: Only needed if multiple apps connected
• --metro-url: Override Metro URL (default: http://localhost:8081)

Errors and warnings include stack trace location. </tool>

<tool name="execute_in_app.py"> Run JavaScript in the app's context (REPL-style).

python scripts/execute_in_app.py "expression" [--no-await] [--timeout 10]

# Or pipe expression
echo "globalThis.myVar" | python scripts/execute_in_app.py

• --no-await: Don't await Promise results
• --app-id: Only needed if multiple apps connected
• --metro-url: Override Metro URL (default: http://localhost:8081)
• Supports any JS: variable access, function calls, IIFEs, etc. </tool>

<tool name="discover_apps.py"> Find React Native apps connected to Metro. **Only use this as a fallback** when `read_logs.py` or `execute_in_app.py` report multiple apps connected.

python scripts/discover_apps.py [--metro-url URL] [--json]

</tool> </tools>

<common_patterns> Check component state:

python scripts/execute_in_app.py "globalThis.__REDUX_DEVTOOLS_EXTENSION__ && globalThis.__REDUX_DEVTOOLS_EXTENSION__.getState()"

Filter for errors:

python scripts/read_logs.py --filter "error|Error|ERROR"

Run diagnostic code:

python scripts/execute_in_app.py "(async () => { const resp = await fetch('http://api.example.com/health'); return resp.status; })()"

</common_patterns>

<success_criteria> Debugging is successful when:

• Connected to the running app via Metro (no connection errors)
• Able to read console logs that help identify the issue
• Able to execute JS to inspect state or run diagnostics
• Root cause identified or issue reproduced </success_criteria>