deep-auditlisted

# Deep Audit - Runtime and Efficiency Investigation Reconstruct how a system actually executes, identify the multipliers that make work operationally significant, and separate proven findings from suspicions. Prefer a few deeply evidenced findings over a catalog of local code smells. ## Scope and Routing Use this skill for repository or subsystem audits centered on: - CPU work and algorithmic growth - allocations, copying, retention, and garbage-collection pressure - filesystem, network, database, process, device, logging, and IPC behavior - concurrency, scheduling, locks, queues, timers, workers, and backpressure - repeated, discarded, eager, overlapping, or retry-amplified work - startup, steady state, failure, cancellation, recovery, and shutdown - latency, throughput, responsiveness, scalability, and resource saturation Route narrower requests elsewhere: - Use `review` for a branch, PR, staged, or working-tree diff review. - Use `diagnosing-bugs` for one reproducible failure or performance regression when the goal includes a fix. - Use `discover` for bounded pre-planning research that does not require a runtime cost model or multi-pass audit state. - Use `qa` for test execution, coverage, and failure triage. - Use a security-specific workflow for authentication, authorization, injection, secrets, supply-chain, or vulnerability assessment. Do not claim security coverage from a runtime-efficiency audit. - Use a provider's orchestration workflow when the primar