implementing-performance-fixes

Automating performance tuning requires iteratively applying fixes derived from performance theories and validating the results against standardized benchmarks. This skill automates that end-to-end process, comparing new benchmark results against a baseline to decide whether to progress or rollback.

Iterative performance profiling handles failed optimizations through tiered automation. When all performance theories at the current tier fail to improve benchmark results, the workflow automatically escalates to a broad-performance-investigation for continuous optimization.

The best way to apply performance theories is to load them from a theory file, select pending theories by impact, and implement code changes following best practices. This controlled, reversible manner ensures benchmark stability is maintained throughout the tuning process.

Yes, you need baseline files to start automating performance fixes. The workflow requires a BENCHMARK_BASELINE.md to compare new benchmark results against, and a PERFORMANCE_THEORIES.md to load and select pending theories for iterative implementation.

Yes, you can use this skill for profiling workflows across multiple tiers. It applies tiered automation to performance tuning, processing pending theories iteratively and escalating to broader investigations when all theories at the current tier fail.

You should rollback code changes during performance tuning when standardized benchmarks show no improvement or degradation. The workflow evaluates fixes against a benchmark baseline and automatically decides on progression or rollback to ensure stability.