What problem does it solve? Static analysis often fails to explain intermittent, timing-dependent, or state-driven Android bugs. This Skill inserts temporary structured trace logs across a suspected call chain to capture runtime evidence, then removes every instrumentation statement after diagnosis so no log code is ever committed. ## Core Features & Use Cases - Call Chain Discovery: Uses CodeGraph exploration to map upstream callers, downstream callees, async boundaries, and lifecycle hooks before instrumenting anything. - Contract-Based Instrumentation: Applies consistent entry/exit/state/exception log conventions with TAG-based filtering, PII masking, and performance guardrails (rate-limiting, counters, single-shot) for high-frequency functions. - Mandatory Cleanup: Mechanically removes all @TRACE- markers after analysis, verifies with assembleDebug and git diff, and refuses to ship instrumentation. - Use Case: A user reports stale data after pull-to-refresh. The Skill traces LoginRepository.authenticate through its coroutine chain, captures logcat output, identifies that a callback arrives after the fragment is destroyed, and produces an OpenSpec-formatted diagnosis. ## Quick Start Trace the call chain starting at LoginRepository.authenticate with temporary logs so I can capture runtime evidence of why the callback fires after the view is destroyed.
