mode-debug

Hypothesis-driven debugging defines testable assumptions and tracks signals to confirm or reject them. This structured approach reveals root causes faster by evaluating evidence across multiple runs instead of relying on guesswork.

To collect NDJSON logs for debugging, insert region markers into your code and run the session to emit logs to .sloppy/debug/debug-<session>.log. Review these structured logs to evaluate your hypotheses and decide next actions.

Use region markers to add focused diagnostic instrumentation in your code. When executed, these markers emit structured NDJSON logs, enabling reproducible analysis across iterative debugging runs to pinpoint exact failure points.

No, this debugging approach requires placing region markers in your code to emit logs to .sloppy/debug. Instrumentation is essential to capture the structured NDJSON logs needed to confirm or reject hypotheses.

The best way to structure a debug loop is to define testable hypotheses, instrument code with region markers, capture NDJSON logs, and interactively refine hypotheses based on user feedback and log evidence.

You should avoid structured logging when you need immediate fixes without code modifications, as this approach requires inserting region markers and generating log files to iteratively evaluate debugging hypotheses.