noyalib-diagnostics-and-toolinglisted

# noyalib diagnostics and tooling The "prove it" skill: the four measurement tools (coverage, Miri, criterion, cargo-fuzz), how to read their output, and the four proof recipes with worked examples from repo history. Triaging a symptom → `noyalib-debugging-playbook`. Deciding what evidence attaches to a PR → `noyalib-validation-and-qa`. This skill is for the layer between: I know what to measure, how do I do it, and what does the output mean. **Extended reference:** the 16-bench inventory, the 10-fuzz-target inventory, the coverage TSV walkthrough, the Miri three-bucket detail, the full `permissive_config()` bench pattern, and the libfuzzer output cheat sheet live in `reference.md` (same directory). Each section below cross-links to its `reference.md` counterpart. ## Provenance (regenerate before trusting) ```sh ls scripts/ # coverage-gap-report, miri, msrv-per-crate, pgo ls crates/noyalib/benches/ # 16 criterion harnesses ls fuzz/fuzz_targets/ # 10 libfuzzer targets ``` Counts at date-stamp: **16** harnesses, **10** fuzz targets. Legacy references to "13 / 9" in `doc/TESTING.md` predate v0.0.14 — the tree is truth. Read for context: `Makefile`, `scripts/miri.sh`, `scripts/coverage-gap-report.sh`, `doc/BENCHMARKS.md`, `doc/TESTING.md`. --- ## 1. Coverage — `cargo +nightly llvm-cov` `scripts/coverage-gap-report.sh` wraps `cargo +nightly llvm-cov --workspace --all-features --no-fail-fast --summary-only