[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79S3C8WDY02DAK0V5VNADQT":3,"skill-same:01M79S3C8WDY02DAK0V5VNADQT":47},{"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":31,"updatedAt":32,"install":33,"resources":46},"01M79S3C8WDY02DAK0V5VNADQT","resharper-clt",null,"Use when running JetBrains ReSharper Command Line Tools over a .NET solution — the CleanupCode\u002Fformat fix pass an agent runs before claiming C# work complete, or the once-per-review-cycle InspectCode gate whose output a code review triages. Do NOT use for non-.NET languages, for judging design or correctness by reading code, as a substitute for the compiler and test gates, or to run InspectCode from inside a per-task completion gate.","dpalfery\u002Fkyber-weave","dpalfery","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F8592256?v=4","01M79RX47GSHMSDA5N4J22JXZ0","David Palfery","01M79RX486WGXFFRB4G5YCSCD5",23,"MIT","https:\u002F\u002Fgithub.com\u002Fdpalfery\u002Fkyber-weave\u002Fblob\u002Fmain\u002Fproducts\u002Fkyber-squad\u002Fskills\u002Fresharper-clt\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: resharper-clt\ndescription: Use when running JetBrains ReSharper Command Line Tools over a .NET solution — the CleanupCode\u002Fformat fix pass an agent runs before claiming C# work complete, or the once-per-review-cycle InspectCode gate whose output a code review triages. Do NOT use for non-.NET languages, for judging design or correctness by reading code, as a substitute for the compiler and test gates, or to run InspectCode from inside a per-task completion gate.\nlicense: MIT\n---\n\n# ReSharper Command Line Tools\n\n`InspectCode` runs ReSharper's inspection set over a solution and emits a machine-readable\nreport. `CleanupCode` applies ReSharper's formatting and cleanup profiles to source files.\nBoth are mechanical: the same solution at the same revision produces the same output, which\nis what makes their result citable as evidence rather than an opinion.\n\nThat property is also the boundary. This skill produces and remediates tool output. It does\nnot judge design, correctness, or intent — those belong to the reviewer and its lenses.\n\n## Setup\n\nThe tools ship as a .NET global tool, pinned in the repository's tool manifest at\n`.config\u002Fdotnet-tools.json`:\n\n```json\n{\n  \"version\": 1,\n  \"isRoot\": true,\n  \"tools\": {\n    \"jetbrains.resharper.globaltools\": {\n      \"version\": \"2026.2.1\",\n      \"commands\": [\"jb\"],\n      \"rollForward\": false\n    }\n  }\n}\n```\n\nALWAYS restore before the first invocation in a session:\n\n```bash\ndotnet tool restore\n```\n\nIf the repository has no manifest entry, say so and stop rather than installing a tool\nversion nobody pinned. A floating analyzer version makes every later report incomparable.\n\n## Commands\n\nResolve the solution path from the repository rather than assuming one, and write reports to\nthe path declared as **.agent_scratchpad** where the repository declares one. Substitute\n`\u003Csolution>` and `\u003Cscratchpad>` below.\n\n| Purpose | Command |\n|---|---|\n| Full-solution inspection | `dotnet jb inspectcode \u003Csolution> --output=\u003Cscratchpad>\u002Finspectcode.xml --format=Xml` |\n| Project-scoped inspection | `dotnet jb inspectcode \u003Csolution> --project=\"\u003CProjectName>\" --output=\u003Cscratchpad>\u002Finspect-\u003Cproject>.xml --format=Xml` |\n| SARIF for CI or automated scanning | `dotnet jb inspectcode \u003Csolution> --output=\u003Cscratchpad>\u002Fresults.sarif --format=Sarif --severity=WARNING` |\n| Formatting and cleanup | `dotnet jb cleanupcode \u003Csolution>` |\n\nScope the run to the affected projects when the solution is large and you are iterating by\nhand. The once-per-review-cycle review gate takes the full solution: it is the only inspection pass in\nthe run, so narrowing it drops coverage nothing else replaces.\n\n## Two cadences, and which one you are in\n\nThis skill has two commands and they run at different times. Getting them the wrong way round is\nthe mistake this section exists to prevent.\n\n| | Command | Who runs it | When |\n|---|---|---|---|\n| **Fix pass** | `dotnet format`, `dotnet format analyzers`, `cleanupcode` | the agent that wrote the code | every task, before `READY_FOR_REVIEW` |\n| **Inspection** | `inspectcode` | the review gate runner | once per review cycle, over the whole accumulated change |\n\nThe reason is contention and cost. `inspectcode` loads the entire solution; run at baseline and\nagain at the end by every worker, it was both the slowest part of a completion gate and the part\nmost likely to collide with a concurrent worker's build output. Run once, on a quiescent tree, it\ncosts a single solution load and reports against a change that is finished. **A worker that runs\n`inspectcode` is doing the council's job at N times the price.**\n\n## Running the fix pass before you claim work is complete\n\nAny agent implementing, modifying, or refactoring C# MUST run this before reporting\n`READY_FOR_REVIEW`.\n\n**Fix mechanically, then stop.** A defect a machine can fix is not a defect worth a reviewer's\nattention. Every mechanical finding that reaches a reviewer costs a review pass to report, a\nrework cycle to fix, and another pass to confirm — to arrive at the same edit `cleanupcode` would\nhave made for free. Erasing them here is the whole point.\n\n1. **Baseline.** Before the first edit, collect language diagnostics for the files you are\n   permitted to change and keep the output. Without it you cannot distinguish a diagnostic you\n   introduced from one that was already there, and \"pre-existing\" is a claim that needs proof.\n   Do **not** take an `inspectcode` baseline; that comparison now belongs to the review gate.\n2. **Fix deterministically, scoped to the files you changed.** Three commands, in this order,\n   each applying rather than verifying:\n\n   ```bash\n   dotnet format \u003Csolution> --include \u003Cchanged files>\n   dotnet format \u003Csolution> analyzers --include \u003Cchanged files>\n   dotnet jb cleanupcode \u003Csolution> --profile=\"\u003Ccleanup profile>\" --include=\"\u003Cchanged files>\" --no-build\n   ```\n\n   `--no-build` skips the tool's own restore and build, so the solution must already be restored\n   into its default `obj\u002F`: run `dotnet restore \u003Csolution>` first when nothing has. A restore\n   redirected with `BaseIntermediateOutputPath` does not count.\n\n   When the host's C# coding standard gives its own cleanupcode command in its `## Completion` section, run that command instead of this line.\n\n   `--include` is not optional. Unscoped, these rewrite files the task never touched, and the\n   change under review stops being the change you made. The pass is idempotent: running it twice\n   produces no second diff, so a re-run after a rework cycle is safe.\n\n   All three load MSBuild and write into `obj\u002F`. Where concurrent workers may run them against the\n   same projects, give each an isolated artifacts path — see the host's per-agent completion gate.\n\n3. **Re-collect diagnostics over the files you changed** — whole file, not only the changed\n   members. Not workspace-wide: while other workers are in flight, a workspace-wide pass reads\n   their half-finished state and attributes it to you.\n4. **Fix every finding in your change set that step 2 could not.** What survives the fix pass is\n   the genuinely non-mechanical remainder, and that is the part worth your judgement.\n5. **Report the result.** The fix pass applied, diagnostics clean on your files, and the isolated\n   artifacts path you used — or an explicit list of what remains with baseline proof.\n\n### What the cleanup profile should and should not do\n\nFix what the repository's own standards and analyzers already require: predefined type keywords,\nexplicit types where the standard forbids `var`, redundant qualifiers and parentheses, unused\nusings, fields that can be `readonly`, formatting. Those are settled decisions, and re-arguing one\nin review is waste.\n\nLeave everything the repository has **not** declared. Brace style with no `.editorconfig` rule\nbehind it, member reordering, and file layout are the tool's opinion rather than the project's\nstandard — and reordering members in particular buries the actual change under churn a reviewer\nthen has to read past. A cleanup profile that imposes undeclared preferences trades review cost\nfor diff noise, which is not the trade this gate exists to make.\n\nA green build is not the same measurement. The compiler and ReSharper's inspection set overlap\npartially and disagree at the edges by design — a suggestion-severity inspection is invisible\nto `dotnet build` and still a real finding. That is why the review gate pins InspectCode at\n`--severity=SUGGESTION` rather than trusting the build.\n\n### Frequently introduced inspections, and what to do\n\n| Inspection | Remediation |\n|---|---|\n| `PossibleMultipleEnumeration` | Materialize the `IEnumerable` with `.ToList()` or `.ToArray()` before iterating it more than once. |\n| `ConditionIsAlwaysTrueOrFalseAccordingToNullableAPIContract` | Fix the contradictory null check, or correct the nullable annotation on the DTO or API model it disagrees with. |\n| `InheritdocInvalidUsage` | Replace `\u003Cinheritdoc \u002F>` on a type that inherits no documentation with a real `\u003Csummary>`. |\n| `UseCollectionExpression` | Modernize the array or list initializer to a collection expression `[...]`. |\n| `CheckNamespace` | Align the file's namespace to its directory hierarchy. |\n| `UnusedParameter.Local`, `PrivateFieldCanBeConvertedToLocalVariable` | Remove the dead parameter, or narrow the single-assignment field to a local. |\n| `RedundantSuppressNullableWarningExpression` | Remove the unnecessary `!`. Suppressing a warning the compiler no longer raises hides the next real one. |\n\n**NEVER fix an inspection by breaking a public contract.** Deleting a public API action\nmethod, an interface member, or a DTO serialization property to silence a \"never used\"\ninspection trades a warning for a breaking change. Suppress it with a stated reason instead,\nor leave it and say why.\n\n## Running it as a review gate\n\nIn a code review this is a **gate**, not a lens: it is a repeatable command whose output the\n`static-analysis-triage` lens attributes to the change. Declare it under `review.gates` in\n`.kyber-weave\u002Fkyber-weave.yml` so the review runner executes it rather than the reviewer\nimprovising a command:\n\n```yaml\nreview:\n  gates:\n    - id: dotnet-tool-restore\n      run: [dotnet, tool, restore]\n      blocking: true\n    - id: inspectcode\n      run: [dotnet, jb, inspectcode, \u003Csolution>, --output=\u003Cscratchpad>\u002Finspectcode.xml, --format=Xml, --severity=SUGGESTION, --caches-home=\u003Cscratchpad>\u002Fgates\u002Finspectcode-cache]\n      blocking: true\n```\n\n`dotnet-tool-restore` provisions the pinned `jb` command; without it the inspection gate fails\non a missing tool rather than on a finding. `--severity=SUGGESTION` states the tool's current\ndefault rather than changing it — pinned, because the suggestion-level inspections are exactly\nthe ones the compiler cannot see, and a default that moves would silently drop them.\n\n`run` is argv, never a command line — the gate runner refuses a shell.\n\nThe gate result records the id, exit code, and duration — not the report. The\n`static-analysis-triage` lens reads the report itself, from the `--output` path the gate\ndeclaration names, so that path is part of the contract between them: change it in one place\nand the lens finds nothing.\n\n**InspectCode exits zero whether or not it finds anything.** `blocking: true` here means the\nanalyzer must have *run*, not that it found nothing. The judgement about what it found is the\nlens's, and it is made against the diff — a solution-wide report is expected to carry\npre-existing findings, and attributing those to the change under review is the specific error\nthat lens exists to avoid.\n\nThe reviewer's own obligations around that output:\n\n- **Do not re-derive it.** The lens attributes diagnostics to the diff and reports them by\n  rule identifier. Reading the code to form a second opinion about a rule is a different\n  lens's job.\n- **Report by rule identifier and location.** \"Analyzer warning\" is not a finding.\n- **A clean run is a recorded result, not silence.** State the gate ran and returned zero\n  findings.\n- **An unavailable tool is a stated skip.** If the manifest has no entry or the restore\n  fails, record the gate as skipped with the reason. A review that quietly omits a gate reads\n  as a review that passed one.\n\n## Worked example\n\nA change adds a service that filters an `IEnumerable\u003CLoad>` and iterates the result twice.\n`dotnet build` is green — nothing here is a compiler error.\n\n**At the task.** The worker runs the fix pass, scoped to its own files and its own build output:\n\n```bash\ndotnet format Hauling.sln --include 2-Application\u002FHauling.Application\u002FDispatch\u002FLoadPlanner.cs \\\n  -p:BaseIntermediateOutputPath=.agent_scratchpad\u002FT3\u002Fartifacts\u002Fobj\u002F\ndotnet format Hauling.sln analyzers --include 2-Application\u002FHauling.Application\u002FDispatch\u002FLoadPlanner.cs \\\n  -p:BaseIntermediateOutputPath=.agent_scratchpad\u002FT3\u002Fartifacts\u002Fobj\u002F\ndotnet restore Hauling.sln\ndotnet jb cleanupcode Hauling.sln --profile=\"\u003Ccleanup profile>\" --include=\"2-Application\u002FHauling.Application\u002FDispatch\u002FLoadPlanner.cs\" --no-build\n```\n\nFormatting, redundant qualifiers, and unused usings are gone. The multiple enumeration is not —\nno formatter can decide that materializing the sequence is the right fix. The worker reports:\n\n```text\nDIAGNOSTICS: clean on 2-Application\u002FHauling.Application\u002FDispatch\u002FLoadPlanner.cs | fix pass: format, format analyzers, cleanupcode — all applied | artifacts: .agent_scratchpad\u002FT3\u002Fartifacts | baseline: .agent_scratchpad\u002FT3\u002Fdiagnostics-baseline.md | remaining: none\n```\n\n**At the end of the run.** The `inspectcode` gate runs once per review cycle over the whole solution:\n\n```xml\n\u003CIssue TypeId=\"PossibleMultipleEnumeration\"\n       File=\"2-Application\u002FHauling.Application\u002FDispatch\u002FLoadPlanner.cs\" Line=\"47\"\n       Message=\"Possible multiple enumeration\" \u002F>\n```\n\n`static-analysis-triage` checks that line against the accumulated diff, finds it inside T3's\nchange, and reports it by rule id and location. It becomes a review finding routed back to a\nworker of that type — `.ToList()` before the first iteration — rather than something every worker\npaid a solution load to look for. Findings on lines the run did not touch are pre-existing and the\nlens drops them; that attribution, not the worker's baseline, is what \"pre-existing\" now means.\n\n## Boundaries\n\n- .NET solutions only. There is no ReSharper CLT for other languages; say so rather than\n  substituting a different tool under this skill's name.\n- `cleanupcode` rewrites source. It belongs to the completion gate above, run by the agent that\n  wrote the code, before any reviewer sees it — and NEVER as part of a review itself. A reviewer\n  that edits the code it is judging has stopped being a reviewer.\n- `inspectcode` is the mirror image: it belongs to the review gate, once per review cycle, and NEVER to a\n  per-task completion gate. A worker that runs it loads the whole solution to analyze a tree its\n  peers are still editing, and pays that cost once per task instead of once per review cycle.\n- Formatter output is not a review finding. Whitespace differences belong to `cleanupcode`, not\n  to the findings list. A reviewer that finds mechanical issues is looking at a change whose\n  completion gate did not run: the correct report is that single fact, not a list of nits the\n  gate would have erased.\n","# ReSharper Command Line Tools\n\n`InspectCode` runs ReSharper's inspection set over a solution and emits a machine-readable\nreport. `CleanupCode` applies ReSharper's formatting and cleanup profiles to source files.\nBoth are mechanical: the same solution at the same revision produces the same output, which\nis what makes their result citable as evidence rather than an opinion.\n\nThat property is also the boundary. This skill produces and remediates tool output. It does\nnot judge design, correctness, or intent — those belong to the reviewer and its lenses.\n\n## Setup\n\nThe tools ship as a .NET global tool, pinned in the repository's tool manifest at\n`.config\u002Fdotnet-tools.json`:\n\n```json\n{\n  \"version\": 1,\n  \"isRoot\": true,\n  \"tools\": {\n    \"jetbrains.resharper.globaltools\": {\n      \"version\": \"2026.2.1\",\n      \"commands\": [\"jb\"],\n      \"rollForward\": false\n    }\n  }\n}\n```\n\nALWAYS restore before the first invocation in a session:\n\n```bash\ndotnet tool restore\n```\n\nIf the repository has no manifest entry, say so and stop rather than installing a tool\nversion nobody pinned. A floating analyzer version makes every later report incomparable.\n\n## Commands\n\nResolve the solution path from the repository rather than assuming one, and write reports to\nthe path declared as **.agent_scratchpad** where the repository declares one. Substitute\n`\u003Csolution>` and `\u003Cscratchpad>` below.\n\n| Purpose | Command |\n|---|---|\n| Full-solution inspection | `dotnet jb inspectcode \u003Csolution> --output=\u003Cscratchpad>\u002Finspectcode.xml --format=Xml` |\n| Project-scoped inspection | `dotnet jb inspectcode \u003Csolution> --project=\"\u003CProjectName>\" --output=\u003Cscratchpad>\u002Finspect-\u003Cproject>.xml --format=Xml` |\n| SARIF for CI or automated scanning | `dotnet jb inspectcode \u003Csolution> --output=\u003Cscratchpad>\u002Fresults.sarif --format=Sarif --severity=WARNING` |\n| Formatting and cleanup | `dotnet jb cleanupcode \u003Csolution>` |\n\nScope the run to the affected projects when the solution is large and you are iterating by\nhand. The once-per-review-cycle review gate takes the full solution: it is the only inspection pass in\nthe run, so narrowing it drops coverage nothing else replaces.\n\n## Two cadences, and which one you are in\n\nThis skill has two commands and they run at different times. Getting them the wrong way round is\nthe mistake this section exists to prevent.\n\n| | Command | Who runs it | When |\n|---|---|---|---|\n| **Fix pass** | `dotnet format`, `dotnet format analyzers`, `cleanupcode` | the agent that wrote the code | every task, before `READY_FOR_REVIEW` |\n| **Inspection** | `inspectcode` | the review gate runner | once per review cycle, over the whole accumulated change |\n\nThe reason is contention and cost. `inspectcode` loads the entire solution; run at baseline and\nagain at the end by every worker, it was both the slowest part of a completion gate and the part\nmost likely to collide with a concurrent worker's build output. Run once, on a quiescent tree, it\ncosts a single solution load and reports against a change that is finished. **A worker that runs\n`inspectcode` is doing the council's job at N times the price.**\n\n## Running the fix pass before you claim work is complete\n\nAny agent implementing, modifying, or refactoring C# MUST run this before reporting\n`READY_FOR_REVIEW`.\n\n**Fix mechanically, then stop.** A defect a machine can fix is not a defect worth a reviewer's\nattention. Every mechanical finding that reaches a reviewer costs a review pass to report, a\nrework cycle to fix, and another pass to confirm — to arrive at the same edit `cleanupcode` would\nhave made for free. Erasing them here is the whole point.\n\n1. **Baseline.** Before the first edit, collect language diagnostics for the files you are\n   permitted to change and keep the output. Without it you cannot distinguish a diagnostic you\n   introduced from one that was already there, and \"pre-existing\" is a claim that needs proof.\n   Do **not** take an `inspectcode` baseline; that comparison now belongs to the review gate.\n2. **Fix deterministically, scoped to the files you changed.** Three commands, in this order,\n   each applying rather than verifying:\n\n   ```bash\n   dotnet format \u003Csolution> --include \u003Cchanged files>\n   dotnet format \u003Csolution> analyzers --include \u003Cchanged files>\n   dotnet jb cleanupcode \u003Csolution> --profile=\"\u003Ccleanup profile>\" --include=\"\u003Cchanged files>\" --no-build\n   ```\n\n   `--no-build` skips the tool's own restore and build, so the solution must already be restored\n   into its default `obj\u002F`: run `dotnet restore \u003Csolution>` first when nothing has. A restore\n   redirected with `BaseIntermediateOutputPath` does not count.\n\n   When the host's C# coding standard gives its own cleanupcode command in its `## Completion` section, run that command instead of this line.\n\n   `--include` is not optional. Unscoped, these rewrite files the task never touched, and the\n   change under review stops being the change you made. The pass is idempotent: running it twice\n   produces no second diff, so a re-run after a rework cycle is safe.\n\n   All three load MSBuild and write into `obj\u002F`. Where concurrent workers may run them against the\n   same projects, give each an isolated artifacts path — see the host's per-agent completion gate.\n\n3. **Re-collect diagnostics over the files you changed** — whole file, not only the changed\n   members. Not workspace-wide: while other workers are in flight, a workspace-wide pass reads\n   their half-finished state and attributes it to you.\n4. **Fix every finding in your change set that step 2 could not.** What survives the fix pass is\n   the genuinely non-mechanical remainder, and that is the part worth your judgement.\n5. **Report the result.** The fix pass applied, diagnostics clean on your files, and the isolated\n   artifacts path you used — or an explicit list of what remains with baseline proof.\n\n### What the cleanup profile should and should not do\n\nFix what the repository's own standards and analyzers already require: predefined type keywords,\nexplicit types where the standard forbids `var`, redundant qualifiers and parentheses, unused\nusings, fields that can be `readonly`, formatting. Those are settled decisions, and re-arguing one\nin review is waste.\n\nLeave everything the repository has **not** declared. Brace style with no `.editorconfig` rule\nbehind it, member reordering, and file layout are the tool's opinion rather than the project's\nstandard — and reordering members in particular buries the actual change under churn a reviewer\nthen has to read past. A cleanup profile that imposes undeclared preferences trades review cost\nfor diff noise, which is not the trade this gate exists to make.\n\nA green build is not the same measurement. The compiler and ReSharper's inspection set overlap\npartially and disagree at the edges by design — a suggestion-severity inspection is invisible\nto `dotnet build` and still a real finding. That is why the review gate pins InspectCode at\n`--severity=SUGGESTION` rather than trusting the build.\n\n### Frequently introduced inspections, and what to do\n\n| Inspection | Remediation |\n|---|---|\n| `PossibleMultipleEnumeration` | Materialize the `IEnumerable` with `.ToList()` or `.ToArray()` before iterating it more than once. |\n| `ConditionIsAlwaysTrueOrFalseAccordingToNullableAPIContract` | Fix the contradictory null check, or correct the nullable annotation on the DTO or API model it disagrees with. |\n| `InheritdocInvalidUsage` | Replace `\u003Cinheritdoc \u002F>` on a type that inherits no documentation with a real `\u003Csummary>`. |\n| `UseCollectionExpression` | Modernize the array or list initializer to a collection expression `[...]`. |\n| `CheckNamespace` | Align the file's namespace to its directory hierarchy. |\n| `UnusedParameter.Local`, `PrivateFieldCanBeConvertedToLocalVariable` | Remove the dead parameter, or narrow the single-assignment field to a local. |\n| `RedundantSuppressNullableWarningExpression` | Remove the unnecessary `!`. Suppressing a warning the compiler no longer raises hides the next real one. |\n\n**NEVER fix an inspection by breaking a public contract.** Deleting a public API action\nmethod, an interface member, or a DTO serialization property to silence a \"never used\"\ninspection trades a warning for a breaking change. Suppress it with a stated reason instead,\nor leave it and say why.\n\n## Running it as a review gate\n\nIn a code review this is a **gate**, not a lens: it is a repeatable command whose output the\n`static-analysis-triage` lens attributes to the change. Declare it under `review.gates` in\n`.kyber-weave\u002Fkyber-weave.yml` so the review runner executes it rather than the reviewer\nimprovising a command:\n\n```yaml\nreview:\n  gates:\n    - id: dotnet-tool-restore\n      run: [dotnet, tool, restore]\n      blocking: true\n    - id: inspectcode\n      run: [dotnet, jb, inspectcode, \u003Csolution>, --output=\u003Cscratchpad>\u002Finspectcode.xml, --format=Xml, --severity=SUGGESTION, --caches-home=\u003Cscratchpad>\u002Fgates\u002Finspectcode-cache]\n      blocking: true\n```\n\n`dotnet-tool-restore` provisions the pinned `jb` command; without it the inspection gate fails\non a missing tool rather than on a finding. `--severity=SUGGESTION` states the tool's current\ndefault rather than changing it — pinned, because the suggestion-level inspections are exactly\nthe ones the compiler cannot see, and a default that moves would silently drop them.\n\n`run` is argv, never a command line — the gate runner refuses a shell.\n\nThe gate result records the id, exit code, and duration — not the report. The\n`static-analysis-triage` lens reads the report itself, from the `--output` path the gate\ndeclaration names, so that path is part of the contract between them: change it in one place\nand the lens finds nothing.\n\n**InspectCode exits zero whether or not it finds anything.** `blocking: true` here means the\nanalyzer must have *run*, not that it found nothing. The judgement about what it found is the\nlens's, and it is made against the diff — a solution-wide report is expected to carry\npre-existing findings, and attributing those to the change under review is the specific error\nthat lens exists to avoid.\n\nThe reviewer's own obligations around that output:\n\n- **Do not re-derive it.** The lens attributes diagnostics to the diff and reports them by\n  rule identifier. Reading the code to form a second opinion about a rule is a different\n  lens's job.\n- **Report by rule identifier and location.** \"Analyzer warning\" is not a finding.\n- **A clean run is a recorded result, not silence.** State the gate ran and returned zero\n  findings.\n- **An unavailable tool is a stated skip.** If the manifest has no entry or the restore\n  fails, record the gate as skipped with the reason. A review that quietly omits a gate reads\n  as a review that passed one.\n\n## Worked example\n\nA change adds a service that filters an `IEnumerable\u003CLoad>` and iterates the result twice.\n`dotnet build` is green — nothing here is a compiler error.\n\n**At the task.** The worker runs the fix pass, scoped to its own files and its own build output:\n\n```bash\ndotnet format Hauling.sln --include 2-Application\u002FHauling.Application\u002FDispatch\u002FLoadPlanner.cs \\\n  -p:BaseIntermediateOutputPath=.agent_scratchpad\u002FT3\u002Fartifacts\u002Fobj\u002F\ndotnet format Hauling.sln analyzers --include 2-Application\u002FHauling.Application\u002FDispatch\u002FLoadPlanner.cs \\\n  -p:BaseIntermediateOutputPath=.agent_scratchpad\u002FT3\u002Fartifacts\u002Fobj\u002F\ndotnet restore Hauling.sln\ndotnet jb cleanupcode Hauling.sln --profile=\"\u003Ccleanup profile>\" --include=\"2-Application\u002FHauling.Application\u002FDispatch\u002FLoadPlanner.cs\" --no-build\n```\n\nFormatting, redundant qualifiers, and unused usings are gone. The multiple enumeration is not —\nno formatter can decide that materializing the sequence is the right fix. The worker reports:\n\n```text\nDIAGNOSTICS: clean on 2-Application\u002FHauling.Application\u002FDispatch\u002FLoadPlanner.cs | fix pass: format, format analyzers, cleanupcode — all applied | artifacts: .agent_scratchpad\u002FT3\u002Fartifacts | baseline: .agent_scratchpad\u002FT3\u002Fdiagnostics-baseline.md | remaining: none\n```\n\n**At the end of the run.** The `inspectcode` gate runs once per review cycle over the whole solution:\n\n```xml\n\u003CIssue TypeId=\"PossibleMultipleEnumeration\"\n       File=\"2-Application\u002FHauling.Application\u002FDispatch\u002FLoadPlanner.cs\" Line=\"47\"\n       Message=\"Possible multiple enumeration\" \u002F>\n```\n\n`static-analysis-triage` checks that line against the accumulated diff, finds it inside T3's\nchange, and reports it by rule id and location. It becomes a review finding routed back to a\nworker of that type — `.ToList()` before the first iteration — rather than something every worker\npaid a solution load to look for. Findings on lines the run did not touch are pre-existing and the\nlens drops them; that attribution, not the worker's baseline, is what \"pre-existing\" now means.\n\n## Boundaries\n\n- .NET solutions only. There is no ReSharper CLT for other languages; say so rather than\n  substituting a different tool under this skill's name.\n- `cleanupcode` rewrites source. It belongs to the completion gate above, run by the agent that\n  wrote the code, before any reviewer sees it — and NEVER as part of a review itself. A reviewer\n  that edits the code it is judging has stopped being a reviewer.\n- `inspectcode` is the mirror image: it belongs to the review gate, once per review cycle, and NEVER to a\n  per-task completion gate. A worker that runs it loads the whole solution to analyze a tree its\n  peers are still editing, and pays that cost once per task instead of once per review cycle.\n- Formatter output is not a review finding. Whitespace differences belong to `cleanupcode`, not\n  to the findings list. A reviewer that finds mechanical issues is looking at a change whose\n  completion gate did not run: the correct report is that single fact, not a list of nits the\n  gate would have erased.\n",{"name":5,"license":15,"description":7},"main","32aee4396c5407e3de876e39eb3c673739175d6b","products\u002Fkyber-squad\u002Fskills\u002Fresharper-clt",3,"https:\u002F\u002Fgithub.com\u002Fdpalfery\u002Fkyber-weave","https:\u002F\u002Fgithub.com\u002Fdpalfery","www.palfery.com","Experienced technology leader with a strong background in cloud strategy, software development, and agile. Passionate about building scalable solutions","2026-11-15 05:04:11",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add dpalfery\u002Fkyber-weave --skill resharper-clt",{"tool":39,"command":40},"bunx","bunx skills add dpalfery\u002Fkyber-weave --skill resharper-clt",{"tool":42,"command":43},"pnpm","pnpm dlx skills add dpalfery\u002Fkyber-weave --skill resharper-clt","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fdpalfery\u002Fkyber-weave\u002Fblob\u002Fmain\u002Fproducts\u002Fkyber-squad\u002Fskills\u002Fresharper-clt\u002FSKILL.md",true,[],[48,55,62,69,76,83],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":19},"01M79S4SA2T3G081B5TBVSF8CV","pr-review","Use to run the pre-merge pull request review checklist so required validations are never skipped. Use when reviewing or approving a pull request before merge. Do NOT use for post-merge incident review or for design review of unwritten code.","https:\u002F\u002Fgithub.com\u002Fdpalfery\u002Fkyber-weave\u002Fblob\u002Fmain\u002Fsamples\u002Fskills\u002Fpr-review\u002FSKILL.md",[],[],{"skillNo":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":59,"categories":60,"keywords":61,"publishedAt":19},"01M79S4PDPV6D0J22CJFDHKBTK","password-reset","Use to reset an employee's corporate account password following the IT security SOP. Use when a user reports they are locked out, forgot their password, or needs a password reset. Do NOT use for multi-factor authentication enrollment, shared\u002Fservice accounts, or external partner accounts.","https:\u002F\u002Fgithub.com\u002Fdpalfery\u002Fkyber-weave\u002Fblob\u002Fmain\u002Fsamples\u002Fskills\u002Fpassword-reset\u002FSKILL.md",[],[],{"skillNo":63,"name":64,"nameZh":6,"description":65,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":66,"categories":67,"keywords":68,"publishedAt":19},"01M79S4K6M0AVKAYA97FKF8R4Y","incident-triage","Use to triage an incoming production incident report, classify its severity, and route it to the right on-call team. Use when someone reports an outage, degradation, or production error. Do NOT use for routine service requests, feature questions, or non-production environments.","https:\u002F\u002Fgithub.com\u002Fdpalfery\u002Fkyber-weave\u002Fblob\u002Fmain\u002Fsamples\u002Fskills\u002Fincident-triage\u002FSKILL.md",[],[],{"skillNo":70,"name":71,"nameZh":6,"description":72,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":73,"categories":74,"keywords":75,"publishedAt":19},"01M79S4GC6M4N8646BRBQJ5WDR","expense-policy","Use as the reference manual for the corporate travel and expense policy covering per-diem limits, approval thresholds, and reimbursable categories. Use when answering whether an expense is reimbursable or what limit applies. Do NOT use to submit, approve, or modify expense reports.","https:\u002F\u002Fgithub.com\u002Fdpalfery\u002Fkyber-weave\u002Fblob\u002Fmain\u002Fsamples\u002Fskills\u002Fexpense-policy\u002FSKILL.md",[],[],{"skillNo":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"publishedAt":19},"01M79S4E2DYDVK71GKZQWQFFCP","Vague_Helper","Helps with HR questions.","https:\u002F\u002Fgithub.com\u002Fdpalfery\u002Fkyber-weave\u002Fblob\u002Fmain\u002Fsamples\u002Fbad-skills\u002Fvague-helper\u002FSKILL.md",[],[],{"skillNo":84,"name":85,"nameZh":6,"description":86,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"publishedAt":19},"01M79S4C13DDHBVF8P99APCSJ4","risky-skill","Use to format a customer record into the standard export template. Use when exporting a single customer record. Do NOT use for bulk exports.","https:\u002F\u002Fgithub.com\u002Fdpalfery\u002Fkyber-weave\u002Fblob\u002Fmain\u002Fsamples\u002Fbad-skills\u002Frisky-skill\u002FSKILL.md",[],[]]