gaia-perf-deepdive

To find N+1 queries and memory bottlenecks, you can use performance analysis tools that scan code stories using deterministic detectors and LLM judgment. This process identifies ORM pattern issues, caching problems, and algorithmic complexity.

Automating performance analysis for code review gates involves running deterministic tools and LLM judgment on your code stories. This emits a machine-readable PASSED or FAILED verdict to ensure performance requirements are met before merging.

LLM judgment detects algorithmic complexity and caching issues by combining deterministic complexity analyzers with semantic reasoning. It evaluates code stories to pinpoint performance bottlenecks that static pattern detectors alone might miss.

Yes, you can use complexity analyzers and memory scanners to check bundle impact. This Skill integrates deterministic tools to evaluate memory and bundle size impacts alongside algorithmic complexity within your code.

Code review fails when performance bottlenecks are not detected because the analysis must emit a machine-readable PASSED or FAILED verdict for the Review Gate. Unresolved N+1 queries or caching issues will trigger a FAILED status.

You need to run a deep dive performance analysis on a code story before merging code to ensure performance requirements are met. It identifies N+1 queries, memory impact, caching issues, and algorithmic complexity.