Menu
Run steelseriesgg-rs's benchmark binaries before and after a change and report the timing delta.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Run the full local CI matrix for steelseriesgg-rs, matching .github/workflows/ci.yml exactly. Use this whenever the user asks to run CI, check all feature variants, run the full test suite, verify before pushing, or run clippy/fmt/tests across features. Invoke proactively before any push or PR review.
Run the protocol fuzzer (`ssgg fuzz`) against a connected keyboard with safety guardrails, capture responses, and fold results back into docs/development/ notes.
Rust conventions for steelseriesgg-rs — error handling, unsafe, feature-gated commands. Use when writing or refactoring any .rs file or Cargo.toml.
Use when user wants to Find and remove genuine dead code using deadcode + vulture. Filters known false positives (dataclass fields, protocol methods, dunder hooks). Run before a cleanup commit.
HID protocol constraints for steelseriesgg-rs device code. Automatically load when editing any file under src/devices/. Contains hard rules about HidReportBuilder usage, unsafe block documentation, and experimental command code labeling.
| name | benchmark-compare |
| description | Run steelseriesgg-rs's benchmark binaries before and after a change and report the timing delta. |
| user-invocable | true |
| disable-model-invocation | true |
Compare performance before/after a change using the three benchmark binaries in src/bin/: benchmark_rgb_alloc, benchmark_fragment, benchmark_validation_io. They print plain timing text to stdout — there is no JSON output or stored baseline, so "before" must be captured from a clean checkout or stash.
src/rgb/ → benchmark_rgb_alloc; touching HID report fragmentation → benchmark_fragment; touching src/validation.rs → benchmark_validation_io). Run all three if unsure or if the change is broad.git stash, run the benchmark(s), save output, git stash pop.cargo run --release --bin benchmark_rgb_alloc
cargo run --release --bin benchmark_fragment
cargo run --release --bin benchmark_validation_io
Instant::now() wall-clock timing with no warmup/statistical repeats beyond their own internal iteration loops — treat single-run deltas as noisy. Re-run at least twice on each side if the result is close to a decision threshold.benchmark_validation_io calls MemorySample::new().await in a loop and can fail mid-run (it reports the iteration count reached before aborting) — a lower iteration count on "after" is itself a regression signal, not just a missing number.