kernel-trace

Trace a feature or subsystem from top to bottom, identify where responsibilities and handoffs live, and recommend durable structural changes. Do not implement code.

Mandatory Protocol

1. Confirm the trace target before analyzing. If the feature or subsystem is unclear, ask for the user-facing flow, starting surface, or relevant files.
2. Read the code from entry point to outcome: UI trigger, validation, transport, business logic, persistence, background work, and return path as applicable.
3. Identify boundaries, handoffs, dependency flow, and failure points across the full path.
4. Report findings with file and line references whenever possible.
5. Separate facts from recommendations.

What To Look For

• Entry points and exit points
• Boundary violations and hidden coupling
• Validation, auth, persistence, and side-effect handoffs
• Circular or overly dense dependencies
• Abstractions that are too thin, too broad, or in the wrong layer
• Missing seams for testing, reuse, observability, or failure recovery