[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M76YHGJEMCJYFPN64WSPDWVG":3,"skill-same:01M76YHGJEMCJYFPN64WSPDWVG":56},{"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":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":30,"ownerDescription":6,"updatedAt":31,"install":32,"resources":45},"01M76YHGJEMCJYFPN64WSPDWVG","test-gap-analysis",null,"Find caller-visible production changes existing tests would miss through pseudo-mutation analysis; verify candidates or close gaps only when requested. Includes read-only per-test composition for grading. Polyglot. USE FOR: would tests catch a bug, behavioral blind spots, surviving mutations. DO NOT USE FOR: concrete fix\u002Ffeature revision proof (test-value-evidence); suite taxonomy (test-tagging); .NET coverage, Cobertura, plateaus or coverage-backed priorities (coverage-analysis); named CRAP (crap-score); new suites (code-testing); assertion\u002Fsmell audits; mutation tools.","stanfish06\u002Fskillquarium","stanfish06","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F93099613?v=4","01M75F1CK4ZW109ZK3T26VN52F","Zhiyuan","01M75F1CM53VH6F79DZ8GXKKES",12,"MIT","https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Ftest-gap-analysis\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: test-gap-analysis\ndescription: >-\n  Find caller-visible production changes existing tests would miss through\n  pseudo-mutation analysis; verify candidates or close gaps only when requested.\n  Includes read-only per-test composition for grading. Polyglot.\n  USE FOR: would tests catch a bug, behavioral blind spots, surviving mutations.\n  DO NOT USE FOR: concrete fix\u002Ffeature revision proof (test-value-evidence);\n  suite taxonomy (test-tagging); .NET coverage, Cobertura, plateaus or\n  coverage-backed priorities (coverage-analysis); named CRAP (crap-score);\n  new suites (code-testing); assertion\u002Fsmell audits; mutation tools.\nlicense: MIT\n---\n\n# Test Gap Analysis\n\nAnswer one question: **which caller-visible production behaviors could change\nwithout an existing test failing?** Mutation reasoning is a probe, not the goal.\nInventory public outcomes first, then verify only credible gaps.\n\n## Composition dispatch\n\nBefore entering the decision flow, check the caller's mode. An explicit\n`per-test-read-only` request (including composition from `grade-tests`) uses\nonly [references\u002Fper-test-read-only.md](references\u002Fper-test-read-only.md).\nThis mode is assessment context conveyed through the host's supported caller\ninstructions, not a new skill-loader parameter or a mode-specific skill name.\nRead that reference and return its per-test assessment inline; **do not enter\nthe standalone workflow below**. Loading this mode does not authorize a\nbaseline run, mutation execution, file edits, suite discovery, or delegation.\nIf the bundled reference is unavailable, allow at most one listing of the known\n`references\u002F` directory, then return **N\u002FA \u002F unverified** with the reason.\n\nRequests without this mode retain the standalone analysis, verification, and\ntest-addition paths below.\n\n## Decision flow\n\n### 1. Set scope\n\nDiscover production and test files from manifests and file types. After a narrow\nsearch misses, inspect the current directory broadly before asking for paths.\n\n| Request | Action |\n|---|---|\n| One component or named risk | Inventory every high-risk public outcome in scope; do not edit production code unless verification was requested |\n| General small-component review | Inventory distinct outcomes and report caller-visible gaps from source\u002Fassertion mapping |\n| Explicit survivor verification | Inventory all requested outcomes; execute one representative observable candidate for each distinct high-risk outcome under verification, then classify it as **Survived** or **Killed** |\n| Explicit exhaustive audit | Read [references\u002Fmutation-catalog.md](references\u002Fmutation-catalog.md) and classify all meaningful candidates |\n| Add tests to an existing suite | Analyze first; add tests only for verified survivors or demonstrated no-coverage outcomes |\n| Create a new suite | Stop and use `code-testing` |\n\nWhen the request names a risk, turn it into a one-line public-outcome allowlist\nbefore reading code. An outcome is not in scope merely because the same method writes it.\nFor `money math`, allow computed or returned amounts, rates, tier\u002Fboundary\nchoice, percentage base\u002Forder, floors\u002Fcaps, and rounding; exclude non-monetary\nstate predicates (including derived booleans), identity, and formatting. Private\ncode is in scope only to trace an allowed outcome.\n\nDo not expand a focused request into a repository audit, plan artifact, or\ndashboard. Use source and tests directly for familiar frameworks. When discovery\nor assertion semantics are unclear, read the matching language file relative\nto the caller-provided or runtime-listed `test-analysis-extensions` catalog;\ndo not invoke the reference-only helper or search installation directories.\nIf absent, state the limitation and use the pinned APIs and rules here.\n\n### 2. Establish one baseline\n\nFor a source-only advisory review, skip execution and label empirical survival\nunverified. A baseline is required when the user requests execution or survivor\nverification, or when closing gaps with tests. Run the narrowest existing test\ncommand once. Choose it from the project\nmanifest; Microsoft.Testing.Platform executables may require `dotnet run`.\nConfirm tests executed: exit 0 with build-only output is not green. If that one\nattempt cannot run the suite, do not troubleshoot the runner or try alternate\ncommands for an advisory review; continue statically and label all candidates\n**unverified**. Do not infer a project-configuration cause from missing output;\nname a cause only when the command reports it.\n\nMissing runner output limits only claims of empirical mutation survival. It does\nnot make source-proven facts tentative: a public outcome with no reaching test\nis still **No coverage**, and an exact expected value derived from the\nunmodified implementation is still actionable. State the baseline limitation\nonce, then give the static source\u002Fassertion conclusion directly instead of\nhedging every row.\n\nFor an advisory review such as \"would tests catch this?\", stop execution after\nthat baseline. Source-to-assertion mapping is sufficient evidence for **No\ncoverage** and **Candidate survivor (unverified)**. Trace or run the unmodified\ncode once only when an original value is unclear. Apply mutations only for\nexplicit verification, an exhaustive audit, or closing gaps with tests.\n\nAny focused mutation budget limits execution, not discovery. Keep every\ndistinct unasserted public outcome in the inventory.\n\n### 3. Inventory public outcomes\n\nFor each public entry point, map:\n\n- input partitions: classifier arms, compound conditions, invalid and\n  nearest-valid guard boundaries, and default cases;\n- each independent observation: returned field\u002Fvariant, exception type,\n  invalid-input acceptance, public state transition, or external side effect;\n- private-helper composition, constants\u002Frates, rounding, retries, cancellation,\n  and error propagation as observed through the public caller.\n\nUse `public input\u002Fsequence -> expected outcome -> existing assertion -> gap`.\nOne asserted return field does not cover another. One allowed result does not\ncover its denial.\n\n**Money math:** inventory the no-op path, every rate\u002Ftier and exact boundary,\noperation order, percentage base or composition, floor\u002Fcap, and rounding. Trace\nprivate helpers through the public result. A test asserting only a broad range\ndoes not pin any exact amount. For each actionable money row, derive one witness\ninput and its exact original result through the complete call chain; do not\nrecommend a generic \"assert the exact amount\" without supplying that amount.\n\n**Ordered guards and retries:** inventory `invalid below minimum | first valid |\nlast allowed or retryable | first blocked | later blocked`. For an upper guard\nsuch as `value >= limit`, use `limit - 1`, `limit`, and `limit + 1`; the last\nwitness exposes narrowing to `value == limit`. Inventory every accepted and\nrejected error class. When type matching is polymorphic, include a representative\nderived accepted type that would expose exact-runtime-type narrowing. A test at\nthe first blocked value does not protect the last allowed or later blocked value.\n\n**Authorization:** enumerate each relevant identity\u002Frole, resource class, and\naction from the caller's view. Untested `false`, forbidden, and unchanged-role\noutcomes are first-class security gaps. Do not analyze variants of an allowed\npath while a denial outcome remains uninventoried. Check each public surface:\n\n- permission-returning APIs: every distinct role\u002Fresource class and every\n  returned capability independently;\n- action-dispatch APIs: each read\u002Fwrite\u002Fdelete-style action branch, especially\n  paths that must return denial;\n- role\u002Fstate transitions: accepted, rejected, invalid, null, and empty inputs,\n  including outcomes that must leave state unchanged.\n\nReserve execution for wholly untested public branches before another variant\nof a partially covered helper. If more than five high-risk behaviors are\nunasserted, execute the top 3-5 and keep the rest visible as **No coverage** or\n**Candidate survivor (unverified)**.\n\nExecution never replaces the ledger. Before mutating or answering, classify\nevery required outcome, including each invalid input, guard boundary,\nclassifier arm, action, and denial.\n\n**Completeness checkpoint:** before selecting findings, explicitly account for\nevery independent mode\u002Fflag, both zero and negative for a `\u003C= 0` guard, every\naccepted exception class, and a representative derived accepted exception when\nmatching is polymorphic. For a removed guard, trace the fallthrough: if it still\nproduces the same public exception type, it is equivalent unless finer exception\nmetadata is an established contract.\n\n### 4. Admit only observable candidates\n\nFirst replay each exact mutation against every existing asserted input or\nsequence with all arguments fixed. Any changed return, exception, state, or side\neffect is **Likely killed**; a dedicated single-purpose test is unnecessary.\nNever compare the mutant on one input with the original on another.\n\nFor survivors, choose a witness before execution or reporting and state\n`witness -> original observation -> mutant observation`. Reuse it in the\nsmallest test. Admit it only when the last two differ publicly after tracing the\nfull call chain; otherwise choose a distinguishing witness or drop it.\n\nExclude:\n\n- edits that require inserting or reordering statements rather than changing or\n  removing an existing expression, condition, constant, return, or side effect;\n- edits that do not compile, including removal of a declaration whose value is\n  still referenced;\n- overflow behavior, exception message\u002F`ParamName` metadata, or other semantics\n  not established by the current contract, source intent, or tests;\n- a removed guard or short-circuit that falls through to the same result,\n  exception, state, and side effects;\n- private representation changes that every public input sequence observes\n  identically, even if the suite stays green;\n- a mutation whose proposed test passes against both original and mutant;\n- boundary edits that return the same value on the distinguishing input; for\n  example, changing `result \u003C floor ? floor : result` to `\u003C=` is equivalent at\n  equality because both branches return `floor`;\n- a standalone auto-property or trivial one-line wrapper\u002Fpredicate with no\n  meaningful branch, calculation, or side effect, unless the user names it;\n- hypothetical future impact, generated code, logging\u002Fformatting-only changes,\n  impossible values, and duplicate syntax variants.\n\nMissing direct assertions do not prove **No coverage**: first trace existing\nassertions through public callers and shared branches. Missing assertions make\nan **observable** candidate a survivor; they do not make an inert mutation\nmeaningful.\n\n### 5. Rank and classify\n\nRank: (1) security denials, financial outcomes, errors, and state changes;\n(2) wholly unasserted public outcomes; (3) boundaries or exact values reached by\nweak assertions; (4) alternate variants of already-asserted behavior.\n\nFinish the inventory before selecting mutations or a verdict. One killed\nattempt, exception type, or switch arm does not clear its siblings.\n\nChoose the verdict from the completed inventory:\n\n- **Strong** when core branches and primary boundaries are protected and only a\n  few validation or default-case variants remain;\n- **Mixed** when meaningful coverage exists but at least one important outcome\n  partition is unprotected;\n- **Weak** when important outcomes are broadly unprotected.\n\nA handful of validation gaps does not make an otherwise broad suite **Mixed**\nunless validation is the named risk or the gaps threaten security, data, or\nother contract-critical behavior.\n\nWhen the inventory meets the **Strong** criteria above, lead with **Strong** and\nname the protected boundaries and dual assertions before listing minor gaps. Do\nnot open with `Mixed`, \"only core paths\", or a risk-heavy dashboard.\n\nStop when existing assertions kill the remaining candidates or no credible\npublic survivor remains. Do not mutate every operator merely to fill a report or\ncalculate a score.\n\n| Result | Meaning |\n|---|---|\n| **Likely killed** | An existing assertion observes the changed outcome |\n| **Candidate survivor (unverified)** | Observable change appears unasserted; not executed |\n| **Killed** | Exact observable mutation executed and a relevant assertion failed |\n| **Survived** | Exact observable mutation executed and tests stayed green |\n| **No coverage** | No test reaches the public outcome; report the missing branch without inventing a survivor |\n| **Equivalent** | No public observation changes; omit from findings |\n\nOutside explicit verification, an exhaustive audit, or a requested test\naddition, execute no mutations. Do not mutate to confirm obvious no coverage.\nFor explicit verification, execute one representative candidate per distinct\nhigh-risk outcome in scope; do not stop after the first one or two while another\nguard, action branch, error class, or denial remains unclassified. Omit\nequivalent syntax variants.\n\n### 6. Verify without creating false positives\n\nEnter this phase only for explicit verification, an exhaustive audit, or a\nrequested test addition.\n\n1. Apply one candidate and confirm the diff changes exactly one intended\n   expression.\n2. Run the narrowest covering test and confirm tests executed: green means\n   **Survived**; a failure at a relevant assertion means **Killed**, for that\n   edit only. Build, discovery, infrastructure, or unrelated failures leave\n   the candidate **unverified**, not Killed.\n3. Revert immediately and confirm the clean source\u002Ftest baseline.\n4. After a green run, re-check the public counterfactual. Execution proves the\n   suite missed the edit, not that the edit changes behavior; drop inert or\n   unobservable mutants.\n\nNever leave mutations in the workspace. Before reporting, reconcile every\nunasserted high-risk outcome as **Survived**, **Candidate survivor\n(unverified)**, **No coverage**, or omitted **Equivalent**. Stop when no credible\npublic gap remains; do not fill a report with internal details or calculate a\nscore unless the user requested an exhaustive audit.\n\n### 7. Close gaps only when requested\n\n1. Add focused tests only for executed **Survived** mutations or demonstrated\n   **No coverage** behavior.\n2. Cover every distinct gap in the requested scope before adding tests\n   for alternate variants of an already-covered behavior.\n3. Before editing, create a survivor-to-test checklist. Before stopping, map\n   every verified survivor to an added test and every added test back to a\n   verified survivor; a passing final suite alone does not prove completeness.\n4. Preserve production code and existing tests when requested.\n5. Prefer one behavior-focused test that kills related mutations over one test\n   per syntax change.\n6. Re-apply the original mutation and prove the new test kills it, then restore\n   the source and run the narrow suite cleanly.\n7. If the fixture or repository supplies a canonical mutation verifier, run\n   that exact command after the tests are added and cite its successful result.\n   Hand-created substitute mutations, a broad green suite, or a test-count\n   increase do not replace the supplied oracle. Once every requested survivor\n   maps to a focused test and the canonical verifier passes, stop; extra tests\n   are not an advantage.\n8. When the request requires existing source or test files to remain unchanged,\n   compare each protected file byte-for-byte with its pre-edit snapshot and\n   report that evidence. Before adding a test, prove its witness differs from\n   every existing case on the relevant branch, boundary, or rounded result so a\n   nominally new test does not duplicate existing coverage.\n\n## Output contract\n\nScale the response to the request.\n\nFor focused or small analysis, return:\n\n1. A one-line verdict: **Strong**, **Mixed**, or **Weak**, with the reason.\n2. For a **Strong** suite, one short strengths sentence naming the concrete\n   protected boundaries, guards, or paired observations that justify the verdict.\n3. One compact row per actionable **Survived**, **Candidate survivor\n   (unverified)**, or **No coverage** outcome. Before adding a row, apply the\n   outcome allowlist when the request names a risk, then apply the\n   observable-candidate rules; omit any candidate that fails either filter.\n   Include every high-risk outcome, use one row per distinct public outcome, and\n   consolidate only related low-risk variants:\n\n   | Risk | Public outcome | Change | Result\u002Fevidence | Smallest test |\n   |---|---|---|---|---|\n\n   Every gap needs a distinguishing witness and a concrete smallest test. An\n   error-path gap must name an invalid input and the expected error\u002Fresult.\n\n4. For a **Mixed** or **Weak** suite, one short strengths sentence naming\n   important killed behavior.\n5. When the request names exclusions, one short scope sentence naming the\n   generated, trivial, or unrelated code intentionally skipped.\n\nDo not repeat the table in prose or report discarded mutants, tool chronology,\nor in-flight reasoning.\n\nFor an exhaustive audit, separate executed **Killed \u002F Survived** counts from\nstatic **Likely killed \u002F Candidate survivor (unverified)** classifications and\n**No coverage \u002F Equivalent** inventory counts. Never include static reasoning\nin an empirical killed\u002Ftotal ratio. Group findings by risk.\n\nFor test additions, name the tests added, the verified mutations they kill, and\nthe successful final command.\n\n## Reliability rules\n\n- A passing test that does not assert the changed outcome does not kill a\n  mutation.\n- Coverage is per behavior partition. One switch\u002Fternary arm or compound input\n  does not prove siblings: allow does not prove deny; read does not prove write;\n  null does not prove empty or whitespace when those inputs have different\n  caller-visible outcomes. A kill clears only the edit and path that ran.\n- Private helpers reached through a public method remain in scope.\n- Error semantics are language-specific: in Rust, `?` propagation versus panic\n  is observable behavior; in C#, exception type and whether an input guard\n  accepts or rejects a value are observable behavior.\n- Cross-check every exact amount or boundary result against the unmodified\n  implementation or an existing exact assertion. If it cannot be checked,\n  state the behavioral relation without inventing a number.\n- Do not label a finding high-risk merely because a mutation survived.\n- Never recommend a redundant test for behavior the existing suite already\n  protects.\n\n## Validation\n\n- [ ] Scope stayed proportional to the request\n- [ ] The original suite passed, or static-only limits are explicit\n- [ ] Every high-risk public outcome in scope was inventoried\n- [ ] Original and mutant have different caller-visible observations\n- [ ] Every outcome labeled **Survived** was executed; unexecuted candidates use\n      **Candidate survivor (unverified)**\n- [ ] Every temporary mutation was reverted\n- [ ] Findings exclude trivial, generated, and equivalent changes\n- [ ] Recommendations target only demonstrated gaps\n- [ ] Every public entry-point branch and each accepted exception type in scope\n      is explicitly accounted for\n- [ ] A supplied canonical mutation verifier was run and reported, not replaced\n      with an ad-hoc proxy\n","# Test Gap Analysis\n\nAnswer one question: **which caller-visible production behaviors could change\nwithout an existing test failing?** Mutation reasoning is a probe, not the goal.\nInventory public outcomes first, then verify only credible gaps.\n\n## Composition dispatch\n\nBefore entering the decision flow, check the caller's mode. An explicit\n`per-test-read-only` request (including composition from `grade-tests`) uses\nonly [references\u002Fper-test-read-only.md](references\u002Fper-test-read-only.md).\nThis mode is assessment context conveyed through the host's supported caller\ninstructions, not a new skill-loader parameter or a mode-specific skill name.\nRead that reference and return its per-test assessment inline; **do not enter\nthe standalone workflow below**. Loading this mode does not authorize a\nbaseline run, mutation execution, file edits, suite discovery, or delegation.\nIf the bundled reference is unavailable, allow at most one listing of the known\n`references\u002F` directory, then return **N\u002FA \u002F unverified** with the reason.\n\nRequests without this mode retain the standalone analysis, verification, and\ntest-addition paths below.\n\n## Decision flow\n\n### 1. Set scope\n\nDiscover production and test files from manifests and file types. After a narrow\nsearch misses, inspect the current directory broadly before asking for paths.\n\n| Request | Action |\n|---|---|\n| One component or named risk | Inventory every high-risk public outcome in scope; do not edit production code unless verification was requested |\n| General small-component review | Inventory distinct outcomes and report caller-visible gaps from source\u002Fassertion mapping |\n| Explicit survivor verification | Inventory all requested outcomes; execute one representative observable candidate for each distinct high-risk outcome under verification, then classify it as **Survived** or **Killed** |\n| Explicit exhaustive audit | Read [references\u002Fmutation-catalog.md](references\u002Fmutation-catalog.md) and classify all meaningful candidates |\n| Add tests to an existing suite | Analyze first; add tests only for verified survivors or demonstrated no-coverage outcomes |\n| Create a new suite | Stop and use `code-testing` |\n\nWhen the request names a risk, turn it into a one-line public-outcome allowlist\nbefore reading code. An outcome is not in scope merely because the same method writes it.\nFor `money math`, allow computed or returned amounts, rates, tier\u002Fboundary\nchoice, percentage base\u002Forder, floors\u002Fcaps, and rounding; exclude non-monetary\nstate predicates (including derived booleans), identity, and formatting. Private\ncode is in scope only to trace an allowed outcome.\n\nDo not expand a focused request into a repository audit, plan artifact, or\ndashboard. Use source and tests directly for familiar frameworks. When discovery\nor assertion semantics are unclear, read the matching language file relative\nto the caller-provided or runtime-listed `test-analysis-extensions` catalog;\ndo not invoke the reference-only helper or search installation directories.\nIf absent, state the limitation and use the pinned APIs and rules here.\n\n### 2. Establish one baseline\n\nFor a source-only advisory review, skip execution and label empirical survival\nunverified. A baseline is required when the user requests execution or survivor\nverification, or when closing gaps with tests. Run the narrowest existing test\ncommand once. Choose it from the project\nmanifest; Microsoft.Testing.Platform executables may require `dotnet run`.\nConfirm tests executed: exit 0 with build-only output is not green. If that one\nattempt cannot run the suite, do not troubleshoot the runner or try alternate\ncommands for an advisory review; continue statically and label all candidates\n**unverified**. Do not infer a project-configuration cause from missing output;\nname a cause only when the command reports it.\n\nMissing runner output limits only claims of empirical mutation survival. It does\nnot make source-proven facts tentative: a public outcome with no reaching test\nis still **No coverage**, and an exact expected value derived from the\nunmodified implementation is still actionable. State the baseline limitation\nonce, then give the static source\u002Fassertion conclusion directly instead of\nhedging every row.\n\nFor an advisory review such as \"would tests catch this?\", stop execution after\nthat baseline. Source-to-assertion mapping is sufficient evidence for **No\ncoverage** and **Candidate survivor (unverified)**. Trace or run the unmodified\ncode once only when an original value is unclear. Apply mutations only for\nexplicit verification, an exhaustive audit, or closing gaps with tests.\n\nAny focused mutation budget limits execution, not discovery. Keep every\ndistinct unasserted public outcome in the inventory.\n\n### 3. Inventory public outcomes\n\nFor each public entry point, map:\n\n- input partitions: classifier arms, compound conditions, invalid and\n  nearest-valid guard boundaries, and default cases;\n- each independent observation: returned field\u002Fvariant, exception type,\n  invalid-input acceptance, public state transition, or external side effect;\n- private-helper composition, constants\u002Frates, rounding, retries, cancellation,\n  and error propagation as observed through the public caller.\n\nUse `public input\u002Fsequence -> expected outcome -> existing assertion -> gap`.\nOne asserted return field does not cover another. One allowed result does not\ncover its denial.\n\n**Money math:** inventory the no-op path, every rate\u002Ftier and exact boundary,\noperation order, percentage base or composition, floor\u002Fcap, and rounding. Trace\nprivate helpers through the public result. A test asserting only a broad range\ndoes not pin any exact amount. For each actionable money row, derive one witness\ninput and its exact original result through the complete call chain; do not\nrecommend a generic \"assert the exact amount\" without supplying that amount.\n\n**Ordered guards and retries:** inventory `invalid below minimum | first valid |\nlast allowed or retryable | first blocked | later blocked`. For an upper guard\nsuch as `value >= limit`, use `limit - 1`, `limit`, and `limit + 1`; the last\nwitness exposes narrowing to `value == limit`. Inventory every accepted and\nrejected error class. When type matching is polymorphic, include a representative\nderived accepted type that would expose exact-runtime-type narrowing. A test at\nthe first blocked value does not protect the last allowed or later blocked value.\n\n**Authorization:** enumerate each relevant identity\u002Frole, resource class, and\naction from the caller's view. Untested `false`, forbidden, and unchanged-role\noutcomes are first-class security gaps. Do not analyze variants of an allowed\npath while a denial outcome remains uninventoried. Check each public surface:\n\n- permission-returning APIs: every distinct role\u002Fresource class and every\n  returned capability independently;\n- action-dispatch APIs: each read\u002Fwrite\u002Fdelete-style action branch, especially\n  paths that must return denial;\n- role\u002Fstate transitions: accepted, rejected, invalid, null, and empty inputs,\n  including outcomes that must leave state unchanged.\n\nReserve execution for wholly untested public branches before another variant\nof a partially covered helper. If more than five high-risk behaviors are\nunasserted, execute the top 3-5 and keep the rest visible as **No coverage** or\n**Candidate survivor (unverified)**.\n\nExecution never replaces the ledger. Before mutating or answering, classify\nevery required outcome, including each invalid input, guard boundary,\nclassifier arm, action, and denial.\n\n**Completeness checkpoint:** before selecting findings, explicitly account for\nevery independent mode\u002Fflag, both zero and negative for a `\u003C= 0` guard, every\naccepted exception class, and a representative derived accepted exception when\nmatching is polymorphic. For a removed guard, trace the fallthrough: if it still\nproduces the same public exception type, it is equivalent unless finer exception\nmetadata is an established contract.\n\n### 4. Admit only observable candidates\n\nFirst replay each exact mutation against every existing asserted input or\nsequence with all arguments fixed. Any changed return, exception, state, or side\neffect is **Likely killed**; a dedicated single-purpose test is unnecessary.\nNever compare the mutant on one input with the original on another.\n\nFor survivors, choose a witness before execution or reporting and state\n`witness -> original observation -> mutant observation`. Reuse it in the\nsmallest test. Admit it only when the last two differ publicly after tracing the\nfull call chain; otherwise choose a distinguishing witness or drop it.\n\nExclude:\n\n- edits that require inserting or reordering statements rather than changing or\n  removing an existing expression, condition, constant, return, or side effect;\n- edits that do not compile, including removal of a declaration whose value is\n  still referenced;\n- overflow behavior, exception message\u002F`ParamName` metadata, or other semantics\n  not established by the current contract, source intent, or tests;\n- a removed guard or short-circuit that falls through to the same result,\n  exception, state, and side effects;\n- private representation changes that every public input sequence observes\n  identically, even if the suite stays green;\n- a mutation whose proposed test passes against both original and mutant;\n- boundary edits that return the same value on the distinguishing input; for\n  example, changing `result \u003C floor ? floor : result` to `\u003C=` is equivalent at\n  equality because both branches return `floor`;\n- a standalone auto-property or trivial one-line wrapper\u002Fpredicate with no\n  meaningful branch, calculation, or side effect, unless the user names it;\n- hypothetical future impact, generated code, logging\u002Fformatting-only changes,\n  impossible values, and duplicate syntax variants.\n\nMissing direct assertions do not prove **No coverage**: first trace existing\nassertions through public callers and shared branches. Missing assertions make\nan **observable** candidate a survivor; they do not make an inert mutation\nmeaningful.\n\n### 5. Rank and classify\n\nRank: (1) security denials, financial outcomes, errors, and state changes;\n(2) wholly unasserted public outcomes; (3) boundaries or exact values reached by\nweak assertions; (4) alternate variants of already-asserted behavior.\n\nFinish the inventory before selecting mutations or a verdict. One killed\nattempt, exception type, or switch arm does not clear its siblings.\n\nChoose the verdict from the completed inventory:\n\n- **Strong** when core branches and primary boundaries are protected and only a\n  few validation or default-case variants remain;\n- **Mixed** when meaningful coverage exists but at least one important outcome\n  partition is unprotected;\n- **Weak** when important outcomes are broadly unprotected.\n\nA handful of validation gaps does not make an otherwise broad suite **Mixed**\nunless validation is the named risk or the gaps threaten security, data, or\nother contract-critical behavior.\n\nWhen the inventory meets the **Strong** criteria above, lead with **Strong** and\nname the protected boundaries and dual assertions before listing minor gaps. Do\nnot open with `Mixed`, \"only core paths\", or a risk-heavy dashboard.\n\nStop when existing assertions kill the remaining candidates or no credible\npublic survivor remains. Do not mutate every operator merely to fill a report or\ncalculate a score.\n\n| Result | Meaning |\n|---|---|\n| **Likely killed** | An existing assertion observes the changed outcome |\n| **Candidate survivor (unverified)** | Observable change appears unasserted; not executed |\n| **Killed** | Exact observable mutation executed and a relevant assertion failed |\n| **Survived** | Exact observable mutation executed and tests stayed green |\n| **No coverage** | No test reaches the public outcome; report the missing branch without inventing a survivor |\n| **Equivalent** | No public observation changes; omit from findings |\n\nOutside explicit verification, an exhaustive audit, or a requested test\naddition, execute no mutations. Do not mutate to confirm obvious no coverage.\nFor explicit verification, execute one representative candidate per distinct\nhigh-risk outcome in scope; do not stop after the first one or two while another\nguard, action branch, error class, or denial remains unclassified. Omit\nequivalent syntax variants.\n\n### 6. Verify without creating false positives\n\nEnter this phase only for explicit verification, an exhaustive audit, or a\nrequested test addition.\n\n1. Apply one candidate and confirm the diff changes exactly one intended\n   expression.\n2. Run the narrowest covering test and confirm tests executed: green means\n   **Survived**; a failure at a relevant assertion means **Killed**, for that\n   edit only. Build, discovery, infrastructure, or unrelated failures leave\n   the candidate **unverified**, not Killed.\n3. Revert immediately and confirm the clean source\u002Ftest baseline.\n4. After a green run, re-check the public counterfactual. Execution proves the\n   suite missed the edit, not that the edit changes behavior; drop inert or\n   unobservable mutants.\n\nNever leave mutations in the workspace. Before reporting, reconcile every\nunasserted high-risk outcome as **Survived**, **Candidate survivor\n(unverified)**, **No coverage**, or omitted **Equivalent**. Stop when no credible\npublic gap remains; do not fill a report with internal details or calculate a\nscore unless the user requested an exhaustive audit.\n\n### 7. Close gaps only when requested\n\n1. Add focused tests only for executed **Survived** mutations or demonstrated\n   **No coverage** behavior.\n2. Cover every distinct gap in the requested scope before adding tests\n   for alternate variants of an already-covered behavior.\n3. Before editing, create a survivor-to-test checklist. Before stopping, map\n   every verified survivor to an added test and every added test back to a\n   verified survivor; a passing final suite alone does not prove completeness.\n4. Preserve production code and existing tests when requested.\n5. Prefer one behavior-focused test that kills related mutations over one test\n   per syntax change.\n6. Re-apply the original mutation and prove the new test kills it, then restore\n   the source and run the narrow suite cleanly.\n7. If the fixture or repository supplies a canonical mutation verifier, run\n   that exact command after the tests are added and cite its successful result.\n   Hand-created substitute mutations, a broad green suite, or a test-count\n   increase do not replace the supplied oracle. Once every requested survivor\n   maps to a focused test and the canonical verifier passes, stop; extra tests\n   are not an advantage.\n8. When the request requires existing source or test files to remain unchanged,\n   compare each protected file byte-for-byte with its pre-edit snapshot and\n   report that evidence. Before adding a test, prove its witness differs from\n   every existing case on the relevant branch, boundary, or rounded result so a\n   nominally new test does not duplicate existing coverage.\n\n## Output contract\n\nScale the response to the request.\n\nFor focused or small analysis, return:\n\n1. A one-line verdict: **Strong**, **Mixed**, or **Weak**, with the reason.\n2. For a **Strong** suite, one short strengths sentence naming the concrete\n   protected boundaries, guards, or paired observations that justify the verdict.\n3. One compact row per actionable **Survived**, **Candidate survivor\n   (unverified)**, or **No coverage** outcome. Before adding a row, apply the\n   outcome allowlist when the request names a risk, then apply the\n   observable-candidate rules; omit any candidate that fails either filter.\n   Include every high-risk outcome, use one row per distinct public outcome, and\n   consolidate only related low-risk variants:\n\n   | Risk | Public outcome | Change | Result\u002Fevidence | Smallest test |\n   |---|---|---|---|---|\n\n   Every gap needs a distinguishing witness and a concrete smallest test. An\n   error-path gap must name an invalid input and the expected error\u002Fresult.\n\n4. For a **Mixed** or **Weak** suite, one short strengths sentence naming\n   important killed behavior.\n5. When the request names exclusions, one short scope sentence naming the\n   generated, trivial, or unrelated code intentionally skipped.\n\nDo not repeat the table in prose or report discarded mutants, tool chronology,\nor in-flight reasoning.\n\nFor an exhaustive audit, separate executed **Killed \u002F Survived** counts from\nstatic **Likely killed \u002F Candidate survivor (unverified)** classifications and\n**No coverage \u002F Equivalent** inventory counts. Never include static reasoning\nin an empirical killed\u002Ftotal ratio. Group findings by risk.\n\nFor test additions, name the tests added, the verified mutations they kill, and\nthe successful final command.\n\n## Reliability rules\n\n- A passing test that does not assert the changed outcome does not kill a\n  mutation.\n- Coverage is per behavior partition. One switch\u002Fternary arm or compound input\n  does not prove siblings: allow does not prove deny; read does not prove write;\n  null does not prove empty or whitespace when those inputs have different\n  caller-visible outcomes. A kill clears only the edit and path that ran.\n- Private helpers reached through a public method remain in scope.\n- Error semantics are language-specific: in Rust, `?` propagation versus panic\n  is observable behavior; in C#, exception type and whether an input guard\n  accepts or rejects a value are observable behavior.\n- Cross-check every exact amount or boundary result against the unmodified\n  implementation or an existing exact assertion. If it cannot be checked,\n  state the behavioral relation without inventing a number.\n- Do not label a finding high-risk merely because a mutation survived.\n- Never recommend a redundant test for behavior the existing suite already\n  protects.\n\n## Validation\n\n- [ ] Scope stayed proportional to the request\n- [ ] The original suite passed, or static-only limits are explicit\n- [ ] Every high-risk public outcome in scope was inventoried\n- [ ] Original and mutant have different caller-visible observations\n- [ ] Every outcome labeled **Survived** was executed; unexecuted candidates use\n      **Candidate survivor (unverified)**\n- [ ] Every temporary mutation was reverted\n- [ ] Findings exclude trivial, generated, and equivalent changes\n- [ ] Recommendations target only demonstrated gaps\n- [ ] Every public entry-point branch and each accepted exception type in scope\n      is explicitly accounted for\n- [ ] A supplied canonical mutation verifier was run and reported, not replaced\n      with an ad-hoc proxy\n",{"name":5,"license":15,"description":7},"master","49e49abf2f670d429e0b7e2c0349739c3fdd5cbe","skills\u002Ftest-gap-analysis",5,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium","https:\u002F\u002Fgithub.com\u002Fstanfish06","https:\u002F\u002Fwww.stanfish.dev\u002F","2026-11-14 21:31:37",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add stanfish06\u002Fskillquarium --skill test-gap-analysis",{"tool":38,"command":39},"bunx","bunx skills add stanfish06\u002Fskillquarium --skill test-gap-analysis",{"tool":41,"command":42},"pnpm","pnpm dlx skills add stanfish06\u002Fskillquarium --skill test-gap-analysis","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Ftest-gap-analysis\u002FSKILL.md",true,[46,52],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"references\u002Fmutation-catalog.md","REFERENCE","参考文档",3942,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Ftest-gap-analysis\u002Freferences\u002Fmutation-catalog.md",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Fper-test-read-only.md",6677,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Ftest-gap-analysis\u002Freferences\u002Fper-test-read-only.md",[57,64,73,80,87,94],{"skillNo":58,"name":59,"nameZh":6,"description":60,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":61,"categories":62,"keywords":63,"publishedAt":19},"01M76ZJTRVA456GD7K8Y3EP82H","zarr-python","Stores and queries chunked N-D scientific arrays with Zarr-Python 3, including codecs, sharding, S3\u002FGCS storage, and NumPy\u002FDask\u002FXarray integration. Use for array layout, bounded I\u002FO, format migration, or scientific metadata preservation.","https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fzarr-python\u002FSKILL.md",[],[],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":68,"licenseFileUrl":69,"sourceHtmlUrl":70,"categories":71,"keywords":72,"publishedAt":19},"01M76ZJQZ15Q1HKC5V85NHE325","yeet","Publish local changes to GitHub by confirming scope, committing intentionally, pushing the branch, and opening a draft PR through the GitHub app from this plugin, with `gh` used only as a fallback where connector coverage is insufficient.","Apache-2.0","https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fyeet\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fyeet\u002FSKILL.md",[],[],{"skillNo":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":68,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":19},"01M76ZJ8ASZ4QYQ70Y6SBA4Z4S","xarray","N-dimensional labeled arrays and datasets in Python. Built on top of NumPy and Dask. It introduces labels in the form of dimensions, coordinates, and attributes on top of raw NumPy-like arrays, making data analysis in physical sciences more intuitive and less error-prone. Use for working with multi-dimensional scientific data, NetCDF\u002FGRIB\u002FZarr files, climate\u002Fweather\u002Foceanographic datasets, remote sensing, geospatial imaging, large out-of-memory datasets with Dask, and labeled array operations.","https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fxarray\u002FSKILL.md",[],[],{"skillNo":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":19},"01M76ZHFBXD6VXPQ02VCEG202B","writing-mstest-tests","Use when asked to fix, rewrite, update, improve, modernize, show corrected code for, or explain existing MSTest tests or MSTest-specific configuration. Use for \"review\" when corrected code or edits are wanted, even for one pasted assertion or passing tests with bad failure output. Covers expected\u002Factual labels; generic Boolean, collection, string, numeric, null, identity, exception, hard-cast, and object[] checks; TestContext\u002Flifecycle; timeout\u002Fcancellation; OS\u002FCI conditions, retry, cleanup, parallelization, MSTest.Sdk project setup, and MSTESTxxxx. Honor the installed MSTest version. DO NOT USE to design new test cases (code-testing), perform report-only audits, create project files rather than explain MSTest setup, run tests, migrate frameworks, or handle non-MSTest\u002Fnon-.NET code.","https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fwriting-mstest-tests\u002FSKILL.md",[],[],{"skillNo":88,"name":89,"nameZh":6,"description":90,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":91,"categories":92,"keywords":93,"publishedAt":19},"01M76ZDFDF4N88RH6HSYGS7GP2","wgs-prs","End-to-end WGS to polygenic risk score pipeline. Takes paired-end FASTQ files (or a pre-existing VCF) through nf-core\u002Fsarek for variant calling, applies VCF QC (normalisation, hard filtering, Ti\u002FTv and Het\u002FHom checks), then computes polygenic risk scores via the PGS Catalog. Fills the FASTQ to VCF gap upstream of the gwas-prs skill.","https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fwgs-prs\u002FSKILL.md",[],[],{"skillNo":95,"name":96,"nameZh":6,"description":97,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":68,"licenseFileUrl":98,"sourceHtmlUrl":99,"categories":100,"keywords":101,"publishedAt":19},"01M76ZCC1SC68J42QK4MDC8RCG","webapp-testing","Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.","https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fwebapp-testing\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fwebapp-testing\u002FSKILL.md",[],[]]