[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78985BTSZ0GD2BY4G74B170":3,"skill-same:01M78985BTSZ0GD2BY4G74B170":67},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":33,"commitSha":34,"skillDirectory":35,"forksCount":36,"repositoryHtmlUrl":37,"repositoryHomepageUrl":38,"ownerHtmlUrl":39,"ownerBlogUrl":40,"ownerDescription":41,"updatedAt":42,"install":43,"resources":56},"01M78985BTSZ0GD2BY4G74B170","noyalib-validation-and-qa",null,"What counts as evidence in noyalib. Use before opening a PR, claiming a fix is done, or adding a test: the evidence bar, test taxonomy, doctest rules, cross-loader parity regression tests, acceptance thresholds. Not running tools (noyalib-diagnostics-and-tooling).","sebastienrousseau\u002Fagtmls","sebastienrousseau","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F1394998?v=4","01M78952QGX0RG3V4QVGESS3N7","Sebastien Rousseau","01M78952R7PPYXEW5VZKQRB0VW",1,"MIT","https:\u002F\u002Fgithub.com\u002Fsebastienrousseau\u002Fagtmls\u002Fblob\u002Fmain\u002Fpacks\u002Fnoyalib\u002Fskills\u002Fnoyalib-validation-and-qa\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: noyalib-validation-and-qa\ndescription: \"What counts as evidence in noyalib. Use before opening a PR, claiming a fix is done, or adding a test: the evidence bar, test taxonomy, doctest rules, cross-loader parity regression tests, acceptance thresholds. Not running tools (noyalib-diagnostics-and-tooling).\"\nlicense: MIT\ncompatibility: \"Requires cargo, git. Tested with Claude Code, Codex, and Aider skill layouts\"\nallowed-tools: \"Read Glob Grep Write Edit Bash WebFetch WebSearch\"\nmetadata:\n  agtmls-owner: \"Sebastien Rousseau\"\n  agtmls-maturity: \"project\"\n  agtmls-bundle: \"noyalib\"\n  agtmls-risk-level: \"medium\"\n  agtmls-network-access: \"optional\"\n  agtmls-writes-files: \"true\"\n  agtmls-executes-commands: \"true\"\n  agtmls-handles-secrets: \"false\"\n  agtmls-requires-human-review: \"true\"\n---\n\n# noyalib Validation & QA\n\nThe evidence bar. A PR that \"looks right\" is not merged; a PR that\nproves a behaviour with a test that would have failed on the previous\ntree is. This skill fences that discipline into checklists.\n\n- Audience: mid-level engineer or Sonnet-class model.\n- Scope: pure-Rust YAML library `noyalib` v0.0.14, test harness and\n  conformance suites.\n- Not this skill:\n  - Running \u002F interpreting the tools → `noyalib-diagnostics-and-tooling`.\n  - The 95 → 98 coverage push specifically → `noyalib-coverage-campaign`.\n  - Chronicle of settled fights → `noyalib-failure-archaeology`.\n- Date-stamped: 2026-07-07.\n\n> **Reference material** (full 130-file test taxonomy, the official\n> YAML Test Suite mechanics + bench-fixture inventory, and the\n> read-only provenance one-liners): see `reference.md` in this\n> directory. This file keeps the evidence bar, the add-a-test\n> checklists, the acceptance thresholds, and what does NOT count.\n\n---\n\n## 1. The evidence bar (non-negotiable)\n\n**A fix is proven by a test that FAILS on the pre-fix tree.** The\ndiscipline is: write the regression test first, watch it fail\n(`cargo test \u003Cname>` red), then fix, then watch green. `CONTRIBUTING.md`\nlines 47–48 codify the rule: *\"Add or update tests in the same commit \u002F\nPR as the behaviour change. Never as a follow-up.\"*\n\nThere is one caveat driven by the [CI-always-green invariant] captured\nin `MEMORY.md`: **red tests never get PUSHED alone.** The red→green\ntransition happens locally. What lands on `main` is the fix plus its\nregression test, both green, in one commit. A reviewer reconstructs the\nred state mentally by reading the test and the fix together — the test\nname typically encodes the pre-fix bug (e.g.\n`from_str_value_refuses_distinct_typed_key_collision`).\n\n**\"Wrong-but-green\" is the enemy.** The silent key-collision collapse\nsurvived 4000+ green tests because no test compared the two loader\npaths against the same input. See\n`tests\u002Fno_span_loader_parity.rs:1–43` — it now pins that any semantics\nshared by two code paths must have a cross-path test. The rule\ngeneralises:\n\n> If a behaviour is implemented by more than one code path\n> (span-full loader vs `NoSpanLoader`, streaming vs blocking,\n> borrowed vs owned, SIMD vs scalar), the invariant needs a test\n> that exercises **both paths against the same input** and asserts\n> they agree.\n\nExemplars to copy:\n\n- `tests\u002Fno_span_loader_parity.rs` — loader parity.\n- `tests\u002Fsimd_equivalence.rs` — SIMD\u002Fscalar path parity.\n- `tests\u002Ferror_kind.rs` — pins the `Error → ErrorKind` classifier\n  (downstream consumers routing on kind).\n\n---\n\n## 2. Test taxonomy\n\n`crates\u002Fnoyalib\u002Ftests\u002F` holds **130 top-level integration test files**\n(`ls crates\u002Fnoyalib\u002Ftests\u002F*.rs | wc -l`), every one SPDX-headed, plus\n**~480 doctests**. When adding a file, match an existing family:\n`coverage_*` (llvm-cov gaps), `cst_*` (green-tree\u002Fround-trip\u002Fspan),\n`spec\u002F\u003Carea>.rs`, `de*`\u002F`ser*`\u002F`serde*` (serde surface), `issue_\u003Cn>`\n(incident regressions), `competitor_bugs.rs`, `proptest.rs`, and the\n`*_panic_regressions.rs` fuzz corpora. Full family-by-family table with\nfile counts in `reference.md` §R1.\n\n---\n\n## 3. Certified \u002F golden inventory\n\nThe conformance gate is `tests\u002Fofficial_suite.rs` — asserts\n`compliance >= 94.0` with an **empty `SKIP_LIST`**; the `README`\n\"387\u002F387 \u002F 19 skipped\" and `BENCHMARKS.md` \"406\u002F406\" numbers are\ndocumentation drift, so cite the test files (351 wrapper `.yaml` cases,\n94% floor), not the prose. Any loaded-case pass-count drop is a release\nblocker. Golden bench fixtures live in `benches\u002Ffixtures\u002F*.yaml` — do\nnot mutate without a paired Criterion baseline. Full mechanics (marker\ndecoder, verbatim assertion arithmetic, companion report, fixture\ntable) in `reference.md` §R2.\n\n---\n\n## 4. How to add a test — checklists per kind\n\n### 4.1 Integration test (new `.rs` file under `tests\u002F`)\n\n1. Pick the family (`coverage_*`, `cst_*`, `de_*`, `issue_\u003Cn>`, spec\u002F,\n   competitive_*, etc.) and match its naming. New file **needs the\n   exact SPDX header** every existing test carries:\n\n   ```rust\n   \u002F\u002F SPDX-License-Identifier: MIT OR Apache-2.0\n   \u002F\u002F Copyright (c) 2026 Noyalib. All rights reserved.\n\n   \u002F\u002F! One-line summary — what invariant this file locks.\n   \u002F\u002F!\n   \u002F\u002F! Longer context: what was wrong, or what path this covers.\n   ```\n\n   (Verify shape against any existing file, e.g. `tests\u002Fissue_46.rs` or\n   `tests\u002Fno_span_loader_parity.rs`.)\n\n2. If it's a bug regression, name the test after the bug and start the\n   file's module doc with `\u002F\u002F! Regression test for …`.\n\n3. If it's spec-area coverage, put it under `tests\u002Fspec\u002F\u003Carea>.rs` and\n   register the module in `tests\u002Fspec\u002Fmod.rs`.\n\n4. Feature-gated tests need either `required-features` in\n   `Cargo.toml`'s `[[test]]` entry, or `#[cfg(feature = \"…\")]` inside\n   the file. Prefer the `cfg` gate — no manifest churn.\n\n5. Anything touching two code paths (blocking vs streaming, spanned\n   vs no-span, SIMD vs scalar) needs a cross-path parity test — see\n   `tests\u002Fno_span_loader_parity.rs` and `tests\u002Fsimd_equivalence.rs`.\n\n### 4.2 Spec case\n\n- One `#[test] fn \u003Csnake_case_name>()` per assertion.\n- File already exists (`tests\u002Fspec\u002Fscalars.rs` etc.) → add the test\n  there. If the area has no file yet, create it under `tests\u002Fspec\u002F`\n  **and** wire it into `tests\u002Fspec\u002Fmod.rs`.\n- Use `noyalib::from_str` (or the specific loader under test) directly;\n  keep the assertion small — one input, one expected value.\n\n### 4.3 Doctest\n\n- Every public item ships a `# Examples` block. `CONTRIBUTING.md:116`\n  makes this explicit.\n- For fallible APIs also add `# Errors`.\n- **Errors-parse-a-repro pattern:** a doctest that demonstrates an\n  error variant must parse an actual reproducer YAML, not just\n  construct the variant. Exemplar (`src\u002Ferror.rs:504–508`):\n\n  ```rust\n  \u002F\u002F\u002F ```\n  \u002F\u002F\u002F use noyalib::{Error, Value, from_str};\n  \u002F\u002F\u002F let err = from_str::\u003CValue>(\"1: a\\n\\\"1\\\": b\\n\").unwrap_err();\n  \u002F\u002F\u002F assert!(matches!(err, Error::KeyCollision(_)));\n  \u002F\u002F\u002F ```\n  ```\n\n  Constructing `Error::KeyCollision(\"1\".into())` on its own is\n  allowed only as a supplementary shape-example; the *behaviour*\n  demonstration must parse.\n\n### 4.4 Property test\n\n- Add to `tests\u002Fproptest.rs` (or a new `properties_\u003Carea>.rs`).\n- The universal property must be readable in one sentence\n  (`parse(emit(v)) == v`, `parse(format(s))` is total, etc.).\n- Shrink to minimal counterexamples; commit those as a fixed regression\n  test in the nearest topical `tests\u002F*.rs` file.\n\n### 4.5 Fuzz-discovered regression\n\n- Add the minimised crashing input to\n  `tests\u002Fscanner_panic_regressions.rs` (or the appropriate\n  `*_panic_regressions.rs`) — one `#[test]` per input.\n- Keep the raw corpus entry under `fuzz\u002Fcorpus\u002F\u003Ctarget>\u002F` so the\n  fuzzer will not re-discover it.\n\n---\n\n## 5. Acceptance thresholds\n\nBefore you claim \"ready to merge\":\n\n| Gate                           | Threshold                                    | Where enforced                                 |\n|--------------------------------|----------------------------------------------|-----------------------------------------------|\n| Official YAML Test Suite       | `compliance >= 94.0` assertion; `SKIP_LIST = &[]` so total-loaded pass count must not drop | `tests\u002Fofficial_suite.rs:295–314` + review |\n| Coverage — functions           | Target 98% \u002F current gate **96%**            | `--fail-under-functions 96` (`ci.yml:160` + `shared-coverage.yml` default) |\n| Coverage — lines               | Target 98% \u002F current gate **94%**            | `--fail-under-lines 94` (`ci.yml:161` + `shared-coverage.yml` default) |\n| Coverage — regions             | Target 98% \u002F current gate **93%**            | `--fail-under-regions 93` (`ci.yml:162` + `shared-coverage.yml` default) |\n| Clippy                         | `-D warnings` — zero warnings                | Workspace CI                                   |\n| Rustfmt                        | Clean (`make fmt`)                           | Workspace CI                                   |\n| Doctests                       | All ~480 pass (`cargo test --doc`)           | Workspace CI                                   |\n| REUSE \u002F SPDX                   | Every source file carries a header           | REUSE workflow                                 |\n| Miri                           | Focused pass green per PR                    | Miri workflow                                  |\n| Fuzz smoke                     | Differential fuzz 10s per target             | `Differential fuzz` workflow                   |\n\nThe 98\u002F98\u002F98 target is the live campaign — see\n`noyalib-coverage-campaign` for the plan to lift the live gate\n(96\u002F94\u002F93 functions\u002Flines\u002Fregions) up to 98\u002F98\u002F98. `doc\u002FTESTING.md`\nstill cites `95 \u002F 93 \u002F 92`; that is documentation drift — trust\nthe workflow files. A conformance regression (any yaml-test-suite\ncase dropping from pass to fail\u002Fskip) is a **release blocker**,\nnot a \"next PR\" item.\n\n---\n\n## 6. What does NOT count as evidence\n\nExplicitly rejected. If any of the below is your entire justification,\nthe PR is not ready:\n\n1. **\"It looks right.\"** — reading the diff is not testing.\n2. **A passing test with no failing-before counterpart.** — the test\n   must be shown (locally or via reviewer's mental replay of the diff)\n   to fail on the pre-fix tree.\n3. **Coverage-only tests that execute but do not assert.** — every\n   `#[test]` asserts something an engineer can read. A test that\n   only calls `from_str` and drops the result is not a test; wrap\n   it with `assert!(matches!(...))` or `assert_eq!` on a value.\n   The `coverage_*` family is not exempt from this rule.\n4. **Bench deltas within noise.** — Criterion prints a p-value; if\n   the CI comparison harness (CodSpeed) does not flag the delta as\n   significant, it is not a regression *and* it is not an\n   improvement. Do not claim either.\n5. **\"Green on my machine.\"** — full CI must be green. See the\n   CI-always-green invariant.\n6. **A doctest that only constructs a type.** — for behavioural\n   claims, parse a reproducer (see §4.3).\n\n---\n\n## 7. When NOT to use this skill\n\n- You need to *run* the tests or interpret llvm-cov output →\n  `noyalib-diagnostics-and-tooling`.\n- You are working the 95 → 98 coverage push specifically →\n  `noyalib-coverage-campaign`.\n- You are researching whether a bug was fixed before →\n  `noyalib-failure-archaeology`.\n- You are deciding *what* to build → `noyalib-architecture-contract`.\n\n---\n\n## Provenance\n\nThe read-only re-verification one-liners (test-file count, SPDX\ncoverage, suite case count, the coded `compliance >=` floor, doctest\ncount, wired coverage thresholds, fixture inventory) live in\n`reference.md` §R3. Run them before quoting any number here.\n\n**Date-stamp:** 2026-07-07 · **Crate version:** v0.0.14 · **Ground\ntruth:** `crates\u002Fnoyalib\u002Ftests\u002F`, `doc\u002FTESTING.md`, `CONTRIBUTING.md`,\n`Makefile`.\n","# noyalib Validation & QA\n\nThe evidence bar. A PR that \"looks right\" is not merged; a PR that\nproves a behaviour with a test that would have failed on the previous\ntree is. This skill fences that discipline into checklists.\n\n- Audience: mid-level engineer or Sonnet-class model.\n- Scope: pure-Rust YAML library `noyalib` v0.0.14, test harness and\n  conformance suites.\n- Not this skill:\n  - Running \u002F interpreting the tools → `noyalib-diagnostics-and-tooling`.\n  - The 95 → 98 coverage push specifically → `noyalib-coverage-campaign`.\n  - Chronicle of settled fights → `noyalib-failure-archaeology`.\n- Date-stamped: 2026-07-07.\n\n> **Reference material** (full 130-file test taxonomy, the official\n> YAML Test Suite mechanics + bench-fixture inventory, and the\n> read-only provenance one-liners): see `reference.md` in this\n> directory. This file keeps the evidence bar, the add-a-test\n> checklists, the acceptance thresholds, and what does NOT count.\n\n---\n\n## 1. The evidence bar (non-negotiable)\n\n**A fix is proven by a test that FAILS on the pre-fix tree.** The\ndiscipline is: write the regression test first, watch it fail\n(`cargo test \u003Cname>` red), then fix, then watch green. `CONTRIBUTING.md`\nlines 47–48 codify the rule: *\"Add or update tests in the same commit \u002F\nPR as the behaviour change. Never as a follow-up.\"*\n\nThere is one caveat driven by the [CI-always-green invariant] captured\nin `MEMORY.md`: **red tests never get PUSHED alone.** The red→green\ntransition happens locally. What lands on `main` is the fix plus its\nregression test, both green, in one commit. A reviewer reconstructs the\nred state mentally by reading the test and the fix together — the test\nname typically encodes the pre-fix bug (e.g.\n`from_str_value_refuses_distinct_typed_key_collision`).\n\n**\"Wrong-but-green\" is the enemy.** The silent key-collision collapse\nsurvived 4000+ green tests because no test compared the two loader\npaths against the same input. See\n`tests\u002Fno_span_loader_parity.rs:1–43` — it now pins that any semantics\nshared by two code paths must have a cross-path test. The rule\ngeneralises:\n\n> If a behaviour is implemented by more than one code path\n> (span-full loader vs `NoSpanLoader`, streaming vs blocking,\n> borrowed vs owned, SIMD vs scalar), the invariant needs a test\n> that exercises **both paths against the same input** and asserts\n> they agree.\n\nExemplars to copy:\n\n- `tests\u002Fno_span_loader_parity.rs` — loader parity.\n- `tests\u002Fsimd_equivalence.rs` — SIMD\u002Fscalar path parity.\n- `tests\u002Ferror_kind.rs` — pins the `Error → ErrorKind` classifier\n  (downstream consumers routing on kind).\n\n---\n\n## 2. Test taxonomy\n\n`crates\u002Fnoyalib\u002Ftests\u002F` holds **130 top-level integration test files**\n(`ls crates\u002Fnoyalib\u002Ftests\u002F*.rs | wc -l`), every one SPDX-headed, plus\n**~480 doctests**. When adding a file, match an existing family:\n`coverage_*` (llvm-cov gaps), `cst_*` (green-tree\u002Fround-trip\u002Fspan),\n`spec\u002F\u003Carea>.rs`, `de*`\u002F`ser*`\u002F`serde*` (serde surface), `issue_\u003Cn>`\n(incident regressions), `competitor_bugs.rs`, `proptest.rs`, and the\n`*_panic_regressions.rs` fuzz corpora. Full family-by-family table with\nfile counts in `reference.md` §R1.\n\n---\n\n## 3. Certified \u002F golden inventory\n\nThe conformance gate is `tests\u002Fofficial_suite.rs` — asserts\n`compliance >= 94.0` with an **empty `SKIP_LIST`**; the `README`\n\"387\u002F387 \u002F 19 skipped\" and `BENCHMARKS.md` \"406\u002F406\" numbers are\ndocumentation drift, so cite the test files (351 wrapper `.yaml` cases,\n94% floor), not the prose. Any loaded-case pass-count drop is a release\nblocker. Golden bench fixtures live in `benches\u002Ffixtures\u002F*.yaml` — do\nnot mutate without a paired Criterion baseline. Full mechanics (marker\ndecoder, verbatim assertion arithmetic, companion report, fixture\ntable) in `reference.md` §R2.\n\n---\n\n## 4. How to add a test — checklists per kind\n\n### 4.1 Integration test (new `.rs` file under `tests\u002F`)\n\n1. Pick the family (`coverage_*`, `cst_*`, `de_*`, `issue_\u003Cn>`, spec\u002F,\n   competitive_*, etc.) and match its naming. New file **needs the\n   exact SPDX header** every existing test carries:\n\n   ```rust\n   \u002F\u002F SPDX-License-Identifier: MIT OR Apache-2.0\n   \u002F\u002F Copyright (c) 2026 Noyalib. All rights reserved.\n\n   \u002F\u002F! One-line summary — what invariant this file locks.\n   \u002F\u002F!\n   \u002F\u002F! Longer context: what was wrong, or what path this covers.\n   ```\n\n   (Verify shape against any existing file, e.g. `tests\u002Fissue_46.rs` or\n   `tests\u002Fno_span_loader_parity.rs`.)\n\n2. If it's a bug regression, name the test after the bug and start the\n   file's module doc with `\u002F\u002F! Regression test for …`.\n\n3. If it's spec-area coverage, put it under `tests\u002Fspec\u002F\u003Carea>.rs` and\n   register the module in `tests\u002Fspec\u002Fmod.rs`.\n\n4. Feature-gated tests need either `required-features` in\n   `Cargo.toml`'s `[[test]]` entry, or `#[cfg(feature = \"…\")]` inside\n   the file. Prefer the `cfg` gate — no manifest churn.\n\n5. Anything touching two code paths (blocking vs streaming, spanned\n   vs no-span, SIMD vs scalar) needs a cross-path parity test — see\n   `tests\u002Fno_span_loader_parity.rs` and `tests\u002Fsimd_equivalence.rs`.\n\n### 4.2 Spec case\n\n- One `#[test] fn \u003Csnake_case_name>()` per assertion.\n- File already exists (`tests\u002Fspec\u002Fscalars.rs` etc.) → add the test\n  there. If the area has no file yet, create it under `tests\u002Fspec\u002F`\n  **and** wire it into `tests\u002Fspec\u002Fmod.rs`.\n- Use `noyalib::from_str` (or the specific loader under test) directly;\n  keep the assertion small — one input, one expected value.\n\n### 4.3 Doctest\n\n- Every public item ships a `# Examples` block. `CONTRIBUTING.md:116`\n  makes this explicit.\n- For fallible APIs also add `# Errors`.\n- **Errors-parse-a-repro pattern:** a doctest that demonstrates an\n  error variant must parse an actual reproducer YAML, not just\n  construct the variant. Exemplar (`src\u002Ferror.rs:504–508`):\n\n  ```rust\n  \u002F\u002F\u002F ```\n  \u002F\u002F\u002F use noyalib::{Error, Value, from_str};\n  \u002F\u002F\u002F let err = from_str::\u003CValue>(\"1: a\\n\\\"1\\\": b\\n\").unwrap_err();\n  \u002F\u002F\u002F assert!(matches!(err, Error::KeyCollision(_)));\n  \u002F\u002F\u002F ```\n  ```\n\n  Constructing `Error::KeyCollision(\"1\".into())` on its own is\n  allowed only as a supplementary shape-example; the *behaviour*\n  demonstration must parse.\n\n### 4.4 Property test\n\n- Add to `tests\u002Fproptest.rs` (or a new `properties_\u003Carea>.rs`).\n- The universal property must be readable in one sentence\n  (`parse(emit(v)) == v`, `parse(format(s))` is total, etc.).\n- Shrink to minimal counterexamples; commit those as a fixed regression\n  test in the nearest topical `tests\u002F*.rs` file.\n\n### 4.5 Fuzz-discovered regression\n\n- Add the minimised crashing input to\n  `tests\u002Fscanner_panic_regressions.rs` (or the appropriate\n  `*_panic_regressions.rs`) — one `#[test]` per input.\n- Keep the raw corpus entry under `fuzz\u002Fcorpus\u002F\u003Ctarget>\u002F` so the\n  fuzzer will not re-discover it.\n\n---\n\n## 5. Acceptance thresholds\n\nBefore you claim \"ready to merge\":\n\n| Gate                           | Threshold                                    | Where enforced                                 |\n|--------------------------------|----------------------------------------------|-----------------------------------------------|\n| Official YAML Test Suite       | `compliance >= 94.0` assertion; `SKIP_LIST = &[]` so total-loaded pass count must not drop | `tests\u002Fofficial_suite.rs:295–314` + review |\n| Coverage — functions           | Target 98% \u002F current gate **96%**            | `--fail-under-functions 96` (`ci.yml:160` + `shared-coverage.yml` default) |\n| Coverage — lines               | Target 98% \u002F current gate **94%**            | `--fail-under-lines 94` (`ci.yml:161` + `shared-coverage.yml` default) |\n| Coverage — regions             | Target 98% \u002F current gate **93%**            | `--fail-under-regions 93` (`ci.yml:162` + `shared-coverage.yml` default) |\n| Clippy                         | `-D warnings` — zero warnings                | Workspace CI                                   |\n| Rustfmt                        | Clean (`make fmt`)                           | Workspace CI                                   |\n| Doctests                       | All ~480 pass (`cargo test --doc`)           | Workspace CI                                   |\n| REUSE \u002F SPDX                   | Every source file carries a header           | REUSE workflow                                 |\n| Miri                           | Focused pass green per PR                    | Miri workflow                                  |\n| Fuzz smoke                     | Differential fuzz 10s per target             | `Differential fuzz` workflow                   |\n\nThe 98\u002F98\u002F98 target is the live campaign — see\n`noyalib-coverage-campaign` for the plan to lift the live gate\n(96\u002F94\u002F93 functions\u002Flines\u002Fregions) up to 98\u002F98\u002F98. `doc\u002FTESTING.md`\nstill cites `95 \u002F 93 \u002F 92`; that is documentation drift — trust\nthe workflow files. A conformance regression (any yaml-test-suite\ncase dropping from pass to fail\u002Fskip) is a **release blocker**,\nnot a \"next PR\" item.\n\n---\n\n## 6. What does NOT count as evidence\n\nExplicitly rejected. If any of the below is your entire justification,\nthe PR is not ready:\n\n1. **\"It looks right.\"** — reading the diff is not testing.\n2. **A passing test with no failing-before counterpart.** — the test\n   must be shown (locally or via reviewer's mental replay of the diff)\n   to fail on the pre-fix tree.\n3. **Coverage-only tests that execute but do not assert.** — every\n   `#[test]` asserts something an engineer can read. A test that\n   only calls `from_str` and drops the result is not a test; wrap\n   it with `assert!(matches!(...))` or `assert_eq!` on a value.\n   The `coverage_*` family is not exempt from this rule.\n4. **Bench deltas within noise.** — Criterion prints a p-value; if\n   the CI comparison harness (CodSpeed) does not flag the delta as\n   significant, it is not a regression *and* it is not an\n   improvement. Do not claim either.\n5. **\"Green on my machine.\"** — full CI must be green. See the\n   CI-always-green invariant.\n6. **A doctest that only constructs a type.** — for behavioural\n   claims, parse a reproducer (see §4.3).\n\n---\n\n## 7. When NOT to use this skill\n\n- You need to *run* the tests or interpret llvm-cov output →\n  `noyalib-diagnostics-and-tooling`.\n- You are working the 95 → 98 coverage push specifically →\n  `noyalib-coverage-campaign`.\n- You are researching whether a bug was fixed before →\n  `noyalib-failure-archaeology`.\n- You are deciding *what* to build → `noyalib-architecture-contract`.\n\n---\n\n## Provenance\n\nThe read-only re-verification one-liners (test-file count, SPDX\ncoverage, suite case count, the coded `compliance >=` floor, doctest\ncount, wired coverage thresholds, fixture inventory) live in\n`reference.md` §R3. Run them before quoting any number here.\n\n**Date-stamp:** 2026-07-07 · **Crate version:** v0.0.14 · **Ground\ntruth:** `crates\u002Fnoyalib\u002Ftests\u002F`, `doc\u002FTESTING.md`, `CONTRIBUTING.md`,\n`Makefile`.\n",{"name":5,"license":15,"metadata":24,"description":7,"allowed-tools":31,"compatibility":32},{"agtmls-owner":12,"agtmls-bundle":25,"agtmls-maturity":26,"agtmls-risk-level":27,"agtmls-writes-files":28,"agtmls-network-access":29,"agtmls-handles-secrets":30,"agtmls-executes-commands":28,"agtmls-requires-human-review":28},"noyalib","project","medium","true","optional","false","Read Glob Grep Write Edit Bash WebFetch WebSearch","Requires cargo, git. Tested with Claude Code, Codex, and Aider skill layouts","main","ecf7dfb9e0505c9584e5838603d8cd9cb0e64efc","packs\u002Fnoyalib\u002Fskills\u002Fnoyalib-validation-and-qa",0,"https:\u002F\u002Fgithub.com\u002Fsebastienrousseau\u002Fagtmls","https:\u002F\u002Fpypi.org\u002Fproject\u002Fagtmls\u002F","https:\u002F\u002Fgithub.com\u002Fsebastienrousseau","https:\u002F\u002Fsebastienrousseau.com","Payments leader, 19 years in banking. I build open-source ISO 20022, post-quantum cryptography and applied AI for finance.","2026-11-09 15:12:45",{"source":8,"skillName":5,"commands":44,"prompt":54,"zipAvailable":55},[45,48,51],{"tool":46,"command":47},"npx","npx skills add sebastienrousseau\u002Fagtmls --skill noyalib-validation-and-qa",{"tool":49,"command":50},"bunx","bunx skills add sebastienrousseau\u002Fagtmls --skill noyalib-validation-and-qa",{"tool":52,"command":53},"pnpm","pnpm dlx skills add sebastienrousseau\u002Fagtmls --skill noyalib-validation-and-qa","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fsebastienrousseau\u002Fagtmls\u002Fblob\u002Fmain\u002Fpacks\u002Fnoyalib\u002Fskills\u002Fnoyalib-validation-and-qa\u002FSKILL.md",true,[57,63],{"path":58,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":61,"sourceHtmlUrl":62},"metadata.json","REFERENCE","参考文档",411,"https:\u002F\u002Fgithub.com\u002Fsebastienrousseau\u002Fagtmls\u002Fblob\u002Fmain\u002Fpacks\u002Fnoyalib\u002Fskills\u002Fnoyalib-validation-and-qa\u002Fmetadata.json",{"path":64,"resourceType":59,"resourceTypeDesc":60,"sizeBytes":65,"sourceHtmlUrl":66},"reference.md",7137,"https:\u002F\u002Fgithub.com\u002Fsebastienrousseau\u002Fagtmls\u002Fblob\u002Fmain\u002Fpacks\u002Fnoyalib\u002Fskills\u002Fnoyalib-validation-and-qa\u002Freference.md",[68,75,82,89,96,103],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":19},"01M789BV3SY53CRXRKQXT13DSB","example-skill","Use when the user needs this reusable workflow. Replace this with trigger-rich wording that says when to load the skill, what task it handles, and when not to use it.","https:\u002F\u002Fgithub.com\u002Fsebastienrousseau\u002Fagtmls\u002Fblob\u002Fmain\u002Ftemplates\u002Fskill\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":19},"01M789BPMXXTXQPSMDQ06DSV1P","writing-plans","Turn a vague or multi-step request into an ordered list of small steps, each with its own done-condition and verification. Load before starting work that touches several files, has unclear scope, or spans more than one session, and when a request needs decomposing before any code is written. Produces a plan a reviewer can check against, not a to-do list.","https:\u002F\u002Fgithub.com\u002Fsebastienrousseau\u002Fagtmls\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-plans\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":19},"01M789BK5BBSDBQENG56CPAEDF","web-research-and-source-triage","Use when gathering current web evidence, comparing sources, triaging credibility, reading GitHub repositories, checking docs, or deciding whether a claim needs live verification.","https:\u002F\u002Fgithub.com\u002Fsebastienrousseau\u002Fagtmls\u002Fblob\u002Fmain\u002Fskills\u002Fweb-research-and-source-triage\u002FSKILL.md",[],[],{"skillNo":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":19},"01M789BF76AAKX8MGWA48YFHMX","vetting-a-skill-before-install","Use before installing, importing or enabling a third-party skill, plugin, marketplace or skills repository in Claude Code, Codex, Antigravity, Cursor or any agent. Pins the source to a commit, audits every skill and the agent configuration it ships, reads what the audit could not cover, and ends in install, mirror or refuse. Not for auditing this registry's own skills.","https:\u002F\u002Fgithub.com\u002Fsebastienrousseau\u002Fagtmls\u002Fblob\u002Fmain\u002Fskills\u002Fvetting-a-skill-before-install\u002FSKILL.md",[],[],{"skillNo":97,"name":98,"nameZh":6,"description":99,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":100,"categories":101,"keywords":102,"publishedAt":19},"01M789B5SQ8G7RCAJBYZ0CR0R4","verification-before-completion","The gate between doing work and claiming it is done. Load before saying finished, fixed, working, or ready, before marking a task complete, and before reporting an outcome to a human. Converts each claim into the observation that would justify it, and names the rationalizations that let unverified work ship. Not the project's own evidence bar or test taxonomy.","https:\u002F\u002Fgithub.com\u002Fsebastienrousseau\u002Fagtmls\u002Fblob\u002Fmain\u002Fskills\u002Fverification-before-completion\u002FSKILL.md",[],[],{"skillNo":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"publishedAt":19},"01M789B1JXCYBRY5P3MWYQBKJ0","using-agtmls","Meta-router for the AgtMLS skill catalog. Read first when unsure which AgtMLS skill applies, when several look relevant, when asking 'is there a skill for porting or noyalib work', or when onboarding to how the hub's skills and shared references fit together.","https:\u002F\u002Fgithub.com\u002Fsebastienrousseau\u002Fagtmls\u002Fblob\u002Fmain\u002Fskills\u002Fusing-agtmls\u002FSKILL.md",[],[]]