[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79AEC4F7WGV5ABHFKX52NFW":3,"skill-same:01M79AEC4F7WGV5ABHFKX52NFW":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":29,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M79AEC4F7WGV5ABHFKX52NFW","nacharbeit-review",null,"Use to run the calibrated judgement review of one or more Claude Code plugins against the nacharbeit rubric: a sonnet finder tuned on planted-defect fixtures until it clears a recall floor for every rule family, measured once on a sealed hold-out, then two-lens finders with gap rounds per (plugin, kind) batch, a haiku routing simulation for cannibalization, an adversarial refuter, and an opus synthesis into a tiered backlog and a findings report. Trigger on 'review this plugin against the Anthropic standard', 'where do these skills and hooks fall short', 'run the nacharbeit review', 'prompt-quality review'. Runs nacharbeit-lint first; never applies a fix (that is nacharbeit-fix). Writes only analysis\u002Fnacharbeit\u002F and, inside werkstoff, the generated findings report.","Anselmoo\u002Fwerkstoff","Anselmoo","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F13209783?v=4","01M799YTB3D7RZ4P3B8CSR2X85","Anselm Hahn","01M799YTBRAM6719D9V0NRGB4R",1,"MIT","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Fplugins\u002Fnacharbeit\u002Fskills\u002Fnacharbeit-review\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: nacharbeit-review\ndescription: \"Use to run the calibrated judgement review of one or more Claude Code plugins against the nacharbeit rubric: a sonnet finder tuned on planted-defect fixtures until it clears a recall floor for every rule family, measured once on a sealed hold-out, then two-lens finders with gap rounds per (plugin, kind) batch, a haiku routing simulation for cannibalization, an adversarial refuter, and an opus synthesis into a tiered backlog and a findings report. Trigger on 'review this plugin against the Anthropic standard', 'where do these skills and hooks fall short', 'run the nacharbeit review', 'prompt-quality review'. Runs nacharbeit-lint first; never applies a fix (that is nacharbeit-fix). Writes only analysis\u002Fnacharbeit\u002F and, inside werkstoff, the generated findings report.\"\n---\n\nMeasure before asserting. The finder that grades real files is the one that first\npassed a planted-defect calibration for every family it will grade; a finder that has\nnot is a reviewer with an unknown false-negative rate, and its output is an opinion.\n\n## Steps\n\n1. **Preflight.** Run `nacharbeit-preflight`. Stop on an open fix lock. Note every\n   skipped checker; the report will carry the same gaps.\n\n2. **Build and bake the args** (this also runs the lint; every path is a flag with a\n   default — `--help` lists them. The calibration fixtures ship inside the plugin and are\n   staged into `analysis\u002Fnacharbeit\u002Ffixtures\u002F`, so this works in any repository):\n\n   ```bash\n   uv run --script \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fbuild_args.py\" [--plugin \u003Cname> ...] [--docs-root docs]\n   ```\n\n   It refuses when a batch kind has no tuning + sealed fixture pair under the fixtures\n   root (`--fixtures-root DIR` swaps the set): a finder cannot grade what it was never\n   calibrated on. It prints the batch\n   list; tell the user the plugin count, the batch count and the fixture count before\n   launching anything.\n\n   `uv run --script` reads the script's PEP 723 header and supplies `pyyaml`. Without `uv`,\n   run the same path with `python3 -B`; that needs `pyyaml` importable, and the linter\n   stops with exit 2 and says so rather than grading frontmatter it cannot parse.\n\n3. **Launch the review** with the Workflow tool, by the baked script path and never\n   with hand-built args — the args are 170+ KB and the leak assertions live in them:\n\n   ```\n   Workflow({ scriptPath: \"analysis\u002Fnacharbeit\u002Frun.js\" })\n   ```\n\n   The workflow calibrates (up to three rounds, an opus rewriter between them), measures\n   the sealed hold-out, refuses to grade if any family falls below its floor, then finds,\n   routes, judges, verifies and synthesizes. Expect on the order of forty agent calls per\n   plugin. A return with `completed: false` names the calls that returned nothing;\n   resume it with `resumeFromRunId`, never persist it.\n\n   *Without the Workflow tool*: dispatch the `component-finder` agent once per batch and\n   lens from the printed batch list, collect its findings, and label every one\n   **uncalibrated** in the report. Do not run `write_results.py` on them — it refuses\n   a return value without a calibration block, and so does the fix pass. An uncalibrated\n   review is a reading, not a measurement, and is presented as one.\n\n4. **Persist, or refuse.** Save the workflow's return value to\n   `analysis\u002Fnacharbeit\u002Freturn.json`, then:\n\n   ```bash\n   python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fwrite_results.py\" analysis\u002Fnacharbeit\u002Freturn.json\n   ```\n\n   It writes `run.json` last, so its presence means the other three are complete. A\n   refusal leaves a `FAILED-\u003Cstamp>` marker and nothing else; report the reason and\n   stop — a stale `run.json` under a fresh marker is exactly the state `build_report.py`\n   refuses to render.\n\n5. **Render the report** and present it:\n\n   ```bash\n   python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fbuild_report.py\"\n   ```\n\n   Inside werkstoff this writes `docs\u002Fprompt-quality-findings.md`; elsewhere pass\n   `--out` and `--no-vitepress`. Lead with section 1 (how much to trust the rest: recall\n   per family, sealed recall, router accuracy), then the per-plugin verdicts, then the\n   backlog by tier with every opus- and human-tier entry named — those are the ones a\n   person must decide.\n\n## Output format\n\n```\nnacharbeit review 2026-09-09T06:12:04Z · rubric 3f9c1a2b7d0e · 3 plugins, 19 batches, 20 fixtures (kinds: assets, docs, hooks, manifest, scripts, skills)\ncalibration: r1 recall 0.76 (min angle 0.40, min family 0.50) → rewrite → r2 recall 0.83 (min angle 0.60, min family 0.67)\nsealed hold-out: 0.79 overall; Q 0.81, HQ 0.75, SQ 0.67, AQ 0.75, PQ 0.80, DQ 1.00 — every family above its floor, finder frozen\nrouter proxy: 0.97 top-1 on 70 known-answer prompts → collisions are measured (24 pairs judged); routing handed to the finders\nrouting findings: 3 Q-ROUTE-MISS, 1 Q-CANN-CAPTURE (produced in code from the simulation)\nraw 312 → verified 164 (+9 from the critic round); declined 5 (fix would make it worse); lint 41\nuncalibrated kinds: agents, commands, workflows — their findings carry calibrated:false\n\n  plugin   verdict      verified  lint  summary\n  andon    needs-work   38        9     andon-loop's argument-hint promises a filter no phase reads; …\n  takt     sound        2         3     …\n  lehre    needs-work   57        8     …\n\nbacklog: haiku 31 · sonnet 96 · opus 2 · human 4\nheld for a person (never auto-applied):\n  [opus ] zirkel: plugins\u002Fzirkel\u002Fworkflows\u002Fsolve.js — design a real gate for each stage's output_contract …\n  [human] andon:   plugins\u002Fandon\u002Fskills\u002Fandon-loop\u002FSKILL.md — decide whether andon-loop supports scoped runs …\n\nreport: docs\u002Fprompt-quality-findings.md (682 lines)\nnext: nacharbeit-fix applies the haiku and sonnet tiers under the lock; the four held entries need you.\n```\n\n## Rules\n\n- **Never retune the finder by hand**, and never edit a fixture after seeing what the\n  finder missed on it. Fix the rubric or add a fixture, then rerun from step 2.\n- **Never pass `args` inline.** The baked `run.js` is the only launch path.\n- **A `completed: false` return is resumed, not persisted.** `write_results.py` would\n  refuse it anyway; do not work around the refusal.\n- **Never grade a kind that has no sealed fixture.** `build_args.py` refuses; do not\n  remove the fixture requirement to make it pass.\n- **Read the transcript of any single-pass surprise** before believing a tally.\n- **Routing below 0.8 is hints, and says so.** The review then emits no `Q-ROUTE-MISS` or\n  `Q-CANN-CAPTURE` and records them under `routing.rulesSkipped`; report that, never a rate.\n- **A `Q-ROUTE-MISS` fix is verified by re-measurement in a fresh process**, never by\n  re-reading the description: definitions load once per session.\n- **Declined findings are not backlog.** Report their count; never hand them to\n  `nacharbeit-fix`.\n- **`calibrated: false` is said aloud.** A finding from a kind with no fixture pair of its own\n  is reported as uncalibrated, not merged into the calibrated tally.\n\n## Resources\n\n- `scripts\u002Fbuild_args.py`, `scripts\u002Fwrite_results.py`, `scripts\u002Fbuild_report.py` — run, in that order.\n- `workflows\u002Freview.js` — read to understand the phases; never edit it mid-run (its baked copy is what runs).\n- `agents\u002Fcomponent-finder.md` — the fallback finder for sessions without the Workflow tool.\n- `references\u002Frubric.md` — the standard every finding cites; read a rule before disputing a finding.\n- `scripts\u002Fambiguous-prompts.json` — werkstoff's hand-written ambiguous prompts for the routing simulation; another repository passes its own with `--ambiguous`.\n- `scripts\u002Ffixtures\u002Fcalibration\u002F` — the planted-defect fixtures the finder is calibrated on; never edited after a finder has seen them.\n","Measure before asserting. The finder that grades real files is the one that first\npassed a planted-defect calibration for every family it will grade; a finder that has\nnot is a reviewer with an unknown false-negative rate, and its output is an opinion.\n\n## Steps\n\n1. **Preflight.** Run `nacharbeit-preflight`. Stop on an open fix lock. Note every\n   skipped checker; the report will carry the same gaps.\n\n2. **Build and bake the args** (this also runs the lint; every path is a flag with a\n   default — `--help` lists them. The calibration fixtures ship inside the plugin and are\n   staged into `analysis\u002Fnacharbeit\u002Ffixtures\u002F`, so this works in any repository):\n\n   ```bash\n   uv run --script \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fbuild_args.py\" [--plugin \u003Cname> ...] [--docs-root docs]\n   ```\n\n   It refuses when a batch kind has no tuning + sealed fixture pair under the fixtures\n   root (`--fixtures-root DIR` swaps the set): a finder cannot grade what it was never\n   calibrated on. It prints the batch\n   list; tell the user the plugin count, the batch count and the fixture count before\n   launching anything.\n\n   `uv run --script` reads the script's PEP 723 header and supplies `pyyaml`. Without `uv`,\n   run the same path with `python3 -B`; that needs `pyyaml` importable, and the linter\n   stops with exit 2 and says so rather than grading frontmatter it cannot parse.\n\n3. **Launch the review** with the Workflow tool, by the baked script path and never\n   with hand-built args — the args are 170+ KB and the leak assertions live in them:\n\n   ```\n   Workflow({ scriptPath: \"analysis\u002Fnacharbeit\u002Frun.js\" })\n   ```\n\n   The workflow calibrates (up to three rounds, an opus rewriter between them), measures\n   the sealed hold-out, refuses to grade if any family falls below its floor, then finds,\n   routes, judges, verifies and synthesizes. Expect on the order of forty agent calls per\n   plugin. A return with `completed: false` names the calls that returned nothing;\n   resume it with `resumeFromRunId`, never persist it.\n\n   *Without the Workflow tool*: dispatch the `component-finder` agent once per batch and\n   lens from the printed batch list, collect its findings, and label every one\n   **uncalibrated** in the report. Do not run `write_results.py` on them — it refuses\n   a return value without a calibration block, and so does the fix pass. An uncalibrated\n   review is a reading, not a measurement, and is presented as one.\n\n4. **Persist, or refuse.** Save the workflow's return value to\n   `analysis\u002Fnacharbeit\u002Freturn.json`, then:\n\n   ```bash\n   python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fwrite_results.py\" analysis\u002Fnacharbeit\u002Freturn.json\n   ```\n\n   It writes `run.json` last, so its presence means the other three are complete. A\n   refusal leaves a `FAILED-\u003Cstamp>` marker and nothing else; report the reason and\n   stop — a stale `run.json` under a fresh marker is exactly the state `build_report.py`\n   refuses to render.\n\n5. **Render the report** and present it:\n\n   ```bash\n   python3 -B \"${CLAUDE_PLUGIN_ROOT}\u002Fscripts\u002Fbuild_report.py\"\n   ```\n\n   Inside werkstoff this writes `docs\u002Fprompt-quality-findings.md`; elsewhere pass\n   `--out` and `--no-vitepress`. Lead with section 1 (how much to trust the rest: recall\n   per family, sealed recall, router accuracy), then the per-plugin verdicts, then the\n   backlog by tier with every opus- and human-tier entry named — those are the ones a\n   person must decide.\n\n## Output format\n\n```\nnacharbeit review 2026-09-09T06:12:04Z · rubric 3f9c1a2b7d0e · 3 plugins, 19 batches, 20 fixtures (kinds: assets, docs, hooks, manifest, scripts, skills)\ncalibration: r1 recall 0.76 (min angle 0.40, min family 0.50) → rewrite → r2 recall 0.83 (min angle 0.60, min family 0.67)\nsealed hold-out: 0.79 overall; Q 0.81, HQ 0.75, SQ 0.67, AQ 0.75, PQ 0.80, DQ 1.00 — every family above its floor, finder frozen\nrouter proxy: 0.97 top-1 on 70 known-answer prompts → collisions are measured (24 pairs judged); routing handed to the finders\nrouting findings: 3 Q-ROUTE-MISS, 1 Q-CANN-CAPTURE (produced in code from the simulation)\nraw 312 → verified 164 (+9 from the critic round); declined 5 (fix would make it worse); lint 41\nuncalibrated kinds: agents, commands, workflows — their findings carry calibrated:false\n\n  plugin   verdict      verified  lint  summary\n  andon    needs-work   38        9     andon-loop's argument-hint promises a filter no phase reads; …\n  takt     sound        2         3     …\n  lehre    needs-work   57        8     …\n\nbacklog: haiku 31 · sonnet 96 · opus 2 · human 4\nheld for a person (never auto-applied):\n  [opus ] zirkel: plugins\u002Fzirkel\u002Fworkflows\u002Fsolve.js — design a real gate for each stage's output_contract …\n  [human] andon:   plugins\u002Fandon\u002Fskills\u002Fandon-loop\u002FSKILL.md — decide whether andon-loop supports scoped runs …\n\nreport: docs\u002Fprompt-quality-findings.md (682 lines)\nnext: nacharbeit-fix applies the haiku and sonnet tiers under the lock; the four held entries need you.\n```\n\n## Rules\n\n- **Never retune the finder by hand**, and never edit a fixture after seeing what the\n  finder missed on it. Fix the rubric or add a fixture, then rerun from step 2.\n- **Never pass `args` inline.** The baked `run.js` is the only launch path.\n- **A `completed: false` return is resumed, not persisted.** `write_results.py` would\n  refuse it anyway; do not work around the refusal.\n- **Never grade a kind that has no sealed fixture.** `build_args.py` refuses; do not\n  remove the fixture requirement to make it pass.\n- **Read the transcript of any single-pass surprise** before believing a tally.\n- **Routing below 0.8 is hints, and says so.** The review then emits no `Q-ROUTE-MISS` or\n  `Q-CANN-CAPTURE` and records them under `routing.rulesSkipped`; report that, never a rate.\n- **A `Q-ROUTE-MISS` fix is verified by re-measurement in a fresh process**, never by\n  re-reading the description: definitions load once per session.\n- **Declined findings are not backlog.** Report their count; never hand them to\n  `nacharbeit-fix`.\n- **`calibrated: false` is said aloud.** A finding from a kind with no fixture pair of its own\n  is reported as uncalibrated, not merged into the calibrated tally.\n\n## Resources\n\n- `scripts\u002Fbuild_args.py`, `scripts\u002Fwrite_results.py`, `scripts\u002Fbuild_report.py` — run, in that order.\n- `workflows\u002Freview.js` — read to understand the phases; never edit it mid-run (its baked copy is what runs).\n- `agents\u002Fcomponent-finder.md` — the fallback finder for sessions without the Workflow tool.\n- `references\u002Frubric.md` — the standard every finding cites; read a rule before disputing a finding.\n- `scripts\u002Fambiguous-prompts.json` — werkstoff's hand-written ambiguous prompts for the routing simulation; another repository passes its own with `--ambiguous`.\n- `scripts\u002Ffixtures\u002Fcalibration\u002F` — the planted-defect fixtures the finder is calibrated on; never edited after a finder has seen them.\n",{"name":5,"description":7},"main","6158c1e4e112c06af2c7ac01b13b2ed7214c0fcf","plugins\u002Fnacharbeit\u002Fskills\u002Fnacharbeit-review",0,"https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff","https:\u002F\u002Fanselmoo.github.io\u002Fwerkstoff\u002F","https:\u002F\u002Fgithub.com\u002FAnselmoo","Likes to Develop","2026-11-02 04:20:59",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add Anselmoo\u002Fwerkstoff --skill nacharbeit-review",{"tool":39,"command":40},"bunx","bunx skills add Anselmoo\u002Fwerkstoff --skill nacharbeit-review",{"tool":42,"command":43},"pnpm","pnpm dlx skills add Anselmoo\u002Fwerkstoff --skill nacharbeit-review","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Fplugins\u002Fnacharbeit\u002Fskills\u002Fnacharbeit-review\u002FSKILL.md",true,[],[48,55,62,67,72,77],{"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},"01M79AJTRBNBX03WWKYJ8ENM45","toy-check","A skill that checks things in the repository and reports what it finds.","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Ftest\u002Fworkflows\u002Ffixtures\u002Ftoy-plugin\u002Fskills\u002Ftoy-check\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},"01M79AJS292F3HTDRXWZVRSN5B","demo-check","Reports which generated files changed since the last generation run. Use when the user asks what drifted. Read-only; never edits a file.","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Ftest\u002Fplugins\u002Ffixtures\u002Fnacharbeit-uncompleted-run\u002Fplugins\u002Fdemo\u002Fskills\u002Fdemo-check\u002FSKILL.md",[],[],{"skillNo":63,"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":64,"categories":65,"keywords":66,"publishedAt":19},"01M79AJP4TFQXKT1AAK1WC1JYK","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Ftest\u002Fplugins\u002Ffixtures\u002Fnacharbeit-stale-lock\u002Fplugins\u002Fdemo\u002Fskills\u002Fdemo-check\u002FSKILL.md",[],[],{"skillNo":68,"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":69,"categories":70,"keywords":71,"publishedAt":19},"01M79AJM6Y50WP3G2BR4H0M60Z","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Ftest\u002Fplugins\u002Ffixtures\u002Fnacharbeit-preflight-mixed\u002Fplugins\u002Fdemo\u002Fskills\u002Fdemo-check\u002FSKILL.md",[],[],{"skillNo":73,"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":74,"categories":75,"keywords":76,"publishedAt":19},"01M79AJGR6DMT8KCCK071V78XM","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Ftest\u002Fplugins\u002Ffixtures\u002Fnacharbeit-hooks-shape\u002Fplugins\u002Fdemo\u002Fskills\u002Fdemo-check\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M79AJBJ4CHC4PZVWKERWF3HF","zirkel-verify-assumptions","Checks exactly ONE named assumption against real evidence in at most 3 steps and reports one outcome: confidence raised (>=90), reading changed, or still unresolved. Use to resolve a single load-bearing uncertainty — typically a blocking entry surfaced by zirkel-clarify-scope: \"verify this one assumption\", \"is it actually true that X\", \"check this before we rely on it\". For multiple uncertainties, invoke once per uncertainty — the step budget is never shared.","https:\u002F\u002Fgithub.com\u002FAnselmoo\u002Fwerkstoff\u002Fblob\u002Fmain\u002Fplugins\u002Fzirkel\u002Fskills\u002Fzirkel-verify-assumptions\u002FSKILL.md",[],[]]