perf-check

name	perf-check
description	Verify no performance regression after code changes — compares before/after benchmarks
disable-model-invocation	true
user-invocable	true
allowed-tools	Bash, Read, Grep, Glob

Performance Regression Check

Verify that code changes haven't introduced performance regressions. This is the gate that must pass before any PR is shipped.

Steps

1. Check for uncommitted changes — run git status to understand what was modified.

2. Identify affected benchmark scenarios by analyzing which files changed:

   • ExpressionBuilder.cs changes → run ALL benchmarks (core compilation)
   • Mapper.cs changes → run FlatMapping, Collection, DeepType benchmarks
   • MapperConfiguration.cs changes → run ALL benchmarks
   • TypeDetails.cs / ReflectionHelper.cs → run ALL benchmarks
   • New feature files only → run the most relevant benchmark + FlatMapping as baseline

3. Run affected benchmarks:

   cd src/EggMapper.Benchmarks && dotnet run -c Release -f net10.0 -- --filter *{scenario}* --exporters json markdown
   

4. Analyze results against constraints:

   • EggMapper MUST be fastest runtime mapper (AutoMapper, Mapster, AgileMapper) on every scenario
   • EggMapper MUST have zero or minimal extra allocations vs manual code
   • Mapperly (source generator) is the only acceptable faster mapper
   • Any regression where EggMapper loses to a runtime competitor is a blocker

5. Report verdict:

   Perf Check: PASS / FAIL

   Changed files: list files Benchmarks run: list scenarios Results: table comparing EggMapper vs competitors

   If FAIL:

   • Which scenario regressed
   • Which competitor is now faster
   • Suggested optimization approach

6. IMPORTANT: Do NOT approve changes that regress performance. Performance is the #1 priority for EggMapper.