[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75GJS1GD5QB3CQE2ENTM82M":3,"skill-same:01M75GJS1GD5QB3CQE2ENTM82M":187},{"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":62,"commitSha":63,"skillDirectory":64,"forksCount":65,"repositoryHtmlUrl":66,"repositoryHomepageUrl":6,"ownerHtmlUrl":67,"ownerBlogUrl":68,"ownerDescription":6,"updatedAt":69,"install":70,"resources":82},"01M75GJS1GD5QB3CQE2ENTM82M","clawpathy-autoresearch",null,"Eval-driven skill tuning. Given a task and an LLM-judge rubric, iteratively rewrites a SKILL.md until a downstream executor agent performs well against the judge. Low-code: all evaluation is LLM-as-judge, not deterministic Python.","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\u002Fclawpathy-autoresearch\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: clawpathy-autoresearch\ndescription: >-\n  Eval-driven skill tuning. Given a task and an LLM-judge rubric, iteratively\n  rewrites a SKILL.md until a downstream executor agent performs well against\n  the judge. Low-code: all evaluation is LLM-as-judge, not deterministic Python.\nversion: 1.0.0\nauthor: Jay Moore\nlicense: MIT\ntags: [meta, autoresearch, skill-tuning, llm-judge, eval-driven]\n\ninputs:\n  - name: paper_query_or_task\n    type: string\n    description: Paper title\u002FURL\u002FPMID\u002FDOI, or a freeform task description\n    required: true\n\noutputs:\n  - name: workspace\u002F\n    type: directory\n    description: Tuned skill\u002FSKILL.md plus history.jsonl, snapshots, executor_runs\n\nmetadata:\n  openclaw:\n    requires:\n      bins: [python3, claude]\n    always: false\n    emoji: \"🔁\"\n    homepage: https:\u002F\u002Fgithub.com\u002FClawBio\u002FClawBio\n    os: [darwin, linux]\n    trigger_keywords:\n      - auto research\n      - autoresearch\n      - tune a skill\n      - skill tuning\n      - improve a skill\n      - eval-driven\n      - clawpathy\n      - replicate paper\n      - reproduce paper\n---\n\n# clawpathy-autoresearch\n\n> [!note] Vault audit 2026-07-24 — USE-10\n> Use this for eval-driven tuning that iteratively rewrites an existing SKILL.md against an LLM-judge rubric; to scaffold a new skill from scratch use `skill-builder`, to auto-draft from observed workflows use `autoskill`, to package a plugin bundle use `plugin-creator`. Distinguishing axis: authoring mode (eval-tuning vs manual scaffold vs observation vs plugin packaging).\n\nEval-driven skill development. The system iteratively rewrites a `SKILL.md`\nso a downstream executor agent performs better at a task class, as judged\nby an LLM against a paper\u002Ftask-specific rubric.\n\n## Core idea\n\n```\n  propose (sonnet)  →  execute (sonnet, shell)  →  judge (opus, rubric)\n       ↑                                                       │\n       └──────── feedback: verdict + recommended edits ────────┘\n```\n\n- **Proposer** rewrites SKILL.md based on the last judge verdict.\n- **Executor** runs the new SKILL.md end-to-end inside a workspace.\n- **Judge** scores methodology (primary) and outputs (secondary) against\n  a per-task rubric. Lower is better; 0 = perfect.\n- Keep the new SKILL.md only if it strictly beats the best score; else\n  revert. Stop on target_score or on `early_stop_n` consecutive regressions.\n\n## You are the orchestrator\n\nYou (the agent reading this) don't run the loop yourself. You dispatch\nsubagents to build the workspace, then hand off to the Python loop.\n\n### Phase 1 — Scout\n\nDispatch a subagent with `prompts\u002Fscout.md` to research the paper\u002Ftask.\nReport key findings to the user in a few lines.\n\n### Phase 2 — Scope (you + user)\n\nHave a conversation. Ask ONE question at a time, multiple-choice where\nhelpful. Agree on:\n- what to reproduce \u002F what success looks like\n- which data sources are in-bounds\n- what methodology expectations belong in the rubric\n- iteration budget and target_score (if any)\n\nPresent a summary and get approval.\n\n### Phase 3 — Build\n\nDispatch a builder subagent with `prompts\u002Fbuilder.md` and the agreed\nscope. It writes:\n- `task.json`\n- `rubric.md` — **the authoritative scoring rubric for the LLM judge**\n- `reference\u002F` (optional; judge-only)\n- `skill\u002FSKILL.md` — seed\n\nValidate:\n```python\nimport sys\nfrom importlib import import_module\nfrom pathlib import Path\n\n# `skills` resolves only from the vault root; the hyphen blocks a plain import.\nsys.path.insert(0, str(Path.home() \u002F \".agents\"))\nvalidate_workspace = import_module(\"skills.clawpathy-autoresearch\").validate_workspace\nprint(validate_workspace(Path(\"WORKSPACE\")))  # [] means valid\n```\n\n### Phase 4 — Loop\n\n```bash\ncd ~\u002F.agents   # run from the vault root, else `No module named 'skills'`\npython -m skills.clawpathy-autoresearch WORKSPACE_DIR\n# or with custom models:\npython -m skills.clawpathy-autoresearch WORKSPACE_DIR \\\n  --proposer-model sonnet --executor-model sonnet --judge-model opus\n```\n\nThe loop streams progress to `WORKSPACE\u002Fhistory.jsonl`, snapshots every\niteration's skill to `WORKSPACE\u002Fsnapshots\u002Fiter-NNN.md`, and writes the\nexecutor's full transcript to `WORKSPACE\u002Fexecutor_runs\u002Fiter-NNN.log`.\n\n## Workspace layout\n\n```\nworkspace\u002F\n  task.json                  # task metadata + loop knobs\n  rubric.md                  # LLM-judge rubric (the heart of the system)\n  reference\u002F                 # optional ground truth, judge-only\n  skill\u002FSKILL.md             # iterated by the loop\n  output\u002F                    # executor outputs (cleared each iter)\n  executor_runs\u002Fiter-NNN.log # transcripts (judge reads these)\n  snapshots\u002Fiter-NNN.md      # per-iter SKILL.md snapshots\n  history.jsonl              # one row per iter: score, kept, verdict\n```\n\n## Key principles\n\n- **LLM judge only.** No deterministic Python scorers. All evaluation goes\n  through `judge.md` + opus. This keeps the system low-code and lets the\n  rubric carry paper-specific nuance without adding code.\n- **Methodology is primary.** The rubric weights \"did the agent use sound\n  methods?\" above \"did the numbers match?\". Ground-truth match is a signal,\n  not the objective — the goal is better SKILL.md files.\n- **Never leak ground truth.** `reference\u002F` is judge-only. The executor\n  prompt says not to read it, and the judge penalises leakage.\n- **No hardcoded answers in SKILL.md.** The proposer prompt and the judge\n  both enforce this. The executor must derive results by running methods.\n- **Snapshots + strict-better revert.** Score on the first iter becomes the\n  floor. Later iters that tie or regress revert to the best.\n\n## Safety\n\n- All processing is local except scout web fetches for public resources.\n- ClawBio disclaimer: research\u002Feducation tool, not a medical device.\n- The executor subagent holds `Bash`, `Write`, and `Edit`, and the instructions it\n  follows are the proposer's output, not human-reviewed text. It runs once per\n  iteration, up to `max_iterations` (default 30), and `--parallel K` runs K at once.\n- Subagents run under `--permission-mode acceptEdits` by default. `--yolo` switches\n  them to `bypassPermissions`, removing every approval prompt including for shell\n  commands. Get the user's explicit consent before passing it, and only for genuinely\n  unattended runs — the Phase 2 approval covers research scope, not this.\n\n## Gotchas\n\n- **Do not skip scoping.** The rubric is paper-specific; a generic rubric\n  tunes nothing. Get the user to agree on methodology expectations.\n- **Do not write a Python scorer.** Earlier versions of this project did.\n  They rewarded API-fetching, not methodology. The judge is the scorer.\n- **Do not hand-pick the \"best\" snapshot yourself.** Trust the loop. If\n  the judge is calibrated wrong, fix the rubric, not the history.\n","# clawpathy-autoresearch\n\n> [!note] Vault audit 2026-07-24 — USE-10\n> Use this for eval-driven tuning that iteratively rewrites an existing SKILL.md against an LLM-judge rubric; to scaffold a new skill from scratch use `skill-builder`, to auto-draft from observed workflows use `autoskill`, to package a plugin bundle use `plugin-creator`. Distinguishing axis: authoring mode (eval-tuning vs manual scaffold vs observation vs plugin packaging).\n\nEval-driven skill development. The system iteratively rewrites a `SKILL.md`\nso a downstream executor agent performs better at a task class, as judged\nby an LLM against a paper\u002Ftask-specific rubric.\n\n## Core idea\n\n```\n  propose (sonnet)  →  execute (sonnet, shell)  →  judge (opus, rubric)\n       ↑                                                       │\n       └──────── feedback: verdict + recommended edits ────────┘\n```\n\n- **Proposer** rewrites SKILL.md based on the last judge verdict.\n- **Executor** runs the new SKILL.md end-to-end inside a workspace.\n- **Judge** scores methodology (primary) and outputs (secondary) against\n  a per-task rubric. Lower is better; 0 = perfect.\n- Keep the new SKILL.md only if it strictly beats the best score; else\n  revert. Stop on target_score or on `early_stop_n` consecutive regressions.\n\n## You are the orchestrator\n\nYou (the agent reading this) don't run the loop yourself. You dispatch\nsubagents to build the workspace, then hand off to the Python loop.\n\n### Phase 1 — Scout\n\nDispatch a subagent with `prompts\u002Fscout.md` to research the paper\u002Ftask.\nReport key findings to the user in a few lines.\n\n### Phase 2 — Scope (you + user)\n\nHave a conversation. Ask ONE question at a time, multiple-choice where\nhelpful. Agree on:\n- what to reproduce \u002F what success looks like\n- which data sources are in-bounds\n- what methodology expectations belong in the rubric\n- iteration budget and target_score (if any)\n\nPresent a summary and get approval.\n\n### Phase 3 — Build\n\nDispatch a builder subagent with `prompts\u002Fbuilder.md` and the agreed\nscope. It writes:\n- `task.json`\n- `rubric.md` — **the authoritative scoring rubric for the LLM judge**\n- `reference\u002F` (optional; judge-only)\n- `skill\u002FSKILL.md` — seed\n\nValidate:\n```python\nimport sys\nfrom importlib import import_module\nfrom pathlib import Path\n\n# `skills` resolves only from the vault root; the hyphen blocks a plain import.\nsys.path.insert(0, str(Path.home() \u002F \".agents\"))\nvalidate_workspace = import_module(\"skills.clawpathy-autoresearch\").validate_workspace\nprint(validate_workspace(Path(\"WORKSPACE\")))  # [] means valid\n```\n\n### Phase 4 — Loop\n\n```bash\ncd ~\u002F.agents   # run from the vault root, else `No module named 'skills'`\npython -m skills.clawpathy-autoresearch WORKSPACE_DIR\n# or with custom models:\npython -m skills.clawpathy-autoresearch WORKSPACE_DIR \\\n  --proposer-model sonnet --executor-model sonnet --judge-model opus\n```\n\nThe loop streams progress to `WORKSPACE\u002Fhistory.jsonl`, snapshots every\niteration's skill to `WORKSPACE\u002Fsnapshots\u002Fiter-NNN.md`, and writes the\nexecutor's full transcript to `WORKSPACE\u002Fexecutor_runs\u002Fiter-NNN.log`.\n\n## Workspace layout\n\n```\nworkspace\u002F\n  task.json                  # task metadata + loop knobs\n  rubric.md                  # LLM-judge rubric (the heart of the system)\n  reference\u002F                 # optional ground truth, judge-only\n  skill\u002FSKILL.md             # iterated by the loop\n  output\u002F                    # executor outputs (cleared each iter)\n  executor_runs\u002Fiter-NNN.log # transcripts (judge reads these)\n  snapshots\u002Fiter-NNN.md      # per-iter SKILL.md snapshots\n  history.jsonl              # one row per iter: score, kept, verdict\n```\n\n## Key principles\n\n- **LLM judge only.** No deterministic Python scorers. All evaluation goes\n  through `judge.md` + opus. This keeps the system low-code and lets the\n  rubric carry paper-specific nuance without adding code.\n- **Methodology is primary.** The rubric weights \"did the agent use sound\n  methods?\" above \"did the numbers match?\". Ground-truth match is a signal,\n  not the objective — the goal is better SKILL.md files.\n- **Never leak ground truth.** `reference\u002F` is judge-only. The executor\n  prompt says not to read it, and the judge penalises leakage.\n- **No hardcoded answers in SKILL.md.** The proposer prompt and the judge\n  both enforce this. The executor must derive results by running methods.\n- **Snapshots + strict-better revert.** Score on the first iter becomes the\n  floor. Later iters that tie or regress revert to the best.\n\n## Safety\n\n- All processing is local except scout web fetches for public resources.\n- ClawBio disclaimer: research\u002Feducation tool, not a medical device.\n- The executor subagent holds `Bash`, `Write`, and `Edit`, and the instructions it\n  follows are the proposer's output, not human-reviewed text. It runs once per\n  iteration, up to `max_iterations` (default 30), and `--parallel K` runs K at once.\n- Subagents run under `--permission-mode acceptEdits` by default. `--yolo` switches\n  them to `bypassPermissions`, removing every approval prompt including for shell\n  commands. Get the user's explicit consent before passing it, and only for genuinely\n  unattended runs — the Phase 2 approval covers research scope, not this.\n\n## Gotchas\n\n- **Do not skip scoping.** The rubric is paper-specific; a generic rubric\n  tunes nothing. Get the user to agree on methodology expectations.\n- **Do not write a Python scorer.** Earlier versions of this project did.\n  They rewarded API-fetching, not methodology. The judge is the scorer.\n- **Do not hand-pick the \"best\" snapshot yourself.** Trust the loop. If\n  the judge is calibrated wrong, fix the rubric, not the history.\n",{"name":5,"tags":24,"author":30,"inputs":31,"license":15,"outputs":37,"version":42,"metadata":43,"description":7},[25,26,27,28,29],"meta","autoresearch","skill-tuning","llm-judge","eval-driven","Jay Moore",[32],{"name":33,"type":34,"required":35,"description":36},"paper_query_or_task","string",true,"Paper title\u002FURL\u002FPMID\u002FDOI, or a freeform task description",[38],{"name":39,"type":40,"description":41},"workspace\u002F","directory","Tuned skill\u002FSKILL.md plus history.jsonl, snapshots, executor_runs","1.0.0",{"openclaw":44},{"os":45,"emoji":48,"always":20,"homepage":49,"requires":50,"trigger_keywords":54},[46,47],"darwin","linux","🔁","https:\u002F\u002Fgithub.com\u002FClawBio\u002FClawBio",{"bins":51},[52,53],"python3","claude",[55,26,56,57,58,29,59,60,61],"auto research","tune a skill","skill tuning","improve a skill","clawpathy","replicate paper","reproduce paper","master","49e49abf2f670d429e0b7e2c0349739c3fdd5cbe","skills\u002Fclawpathy-autoresearch",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":71,"prompt":81,"zipAvailable":35},[72,75,78],{"tool":73,"command":74},"npx","npx skills add stanfish06\u002Fskillquarium --skill clawpathy-autoresearch",{"tool":76,"command":77},"bunx","bunx skills add stanfish06\u002Fskillquarium --skill clawpathy-autoresearch",{"tool":79,"command":80},"pnpm","pnpm dlx skills add stanfish06\u002Fskillquarium --skill clawpathy-autoresearch","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002FSKILL.md",[83,89,93,99,103,109,113,117,121,125,129,133,139,143,147,151,155,159,163,167,171,175,179,183],{"path":84,"resourceType":85,"resourceTypeDesc":86,"sizeBytes":87,"sourceHtmlUrl":88},"__init__.py","SCRIPT","脚本",383,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002F__init__.py",{"path":90,"resourceType":85,"resourceTypeDesc":86,"sizeBytes":91,"sourceHtmlUrl":92},"__main__.py",1857,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002F__main__.py",{"path":94,"resourceType":95,"resourceTypeDesc":96,"sizeBytes":97,"sourceHtmlUrl":98},".gitignore","OTHER","其他",957,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002F.gitignore",{"path":100,"resourceType":85,"resourceTypeDesc":86,"sizeBytes":101,"sourceHtmlUrl":102},"dispatcher.py",2815,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fdispatcher.py",{"path":104,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":107,"sourceHtmlUrl":108},"examples\u002Fchampions\u002FREADME.md","REFERENCE","参考文档",1397,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fexamples\u002Fchampions\u002FREADME.md",{"path":110,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":111,"sourceHtmlUrl":112},"examples\u002Fchampions\u002Ftrubetskoy_scz_finemap_0.235.md",50954,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fexamples\u002Fchampions\u002Ftrubetskoy_scz_finemap_0.235.md",{"path":114,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":115,"sourceHtmlUrl":116},"examples\u002Fchampions\u002Fyengo_height_ldsc_h2.md",26312,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fexamples\u002Fchampions\u002Fyengo_height_ldsc_h2.md",{"path":118,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":119,"sourceHtmlUrl":120},"examples\u002Fdemo_task\u002FREADME.md",690,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fexamples\u002Fdemo_task\u002FREADME.md",{"path":122,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":123,"sourceHtmlUrl":124},"examples\u002Fdemo_task\u002Frubric.md",1763,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fexamples\u002Fdemo_task\u002Frubric.md",{"path":126,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":127,"sourceHtmlUrl":128},"examples\u002Fdemo_task\u002Ftask.json",463,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fexamples\u002Fdemo_task\u002Ftask.json",{"path":130,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":131,"sourceHtmlUrl":132},"examples\u002Fpresentation_intro.md",3637,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fexamples\u002Fpresentation_intro.md",{"path":134,"resourceType":135,"resourceTypeDesc":136,"sizeBytes":137,"sourceHtmlUrl":138},"examples\u002Fskill_transfer_summary.png","ASSET","资产文件",120434,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fexamples\u002Fskill_transfer_summary.png",{"path":140,"resourceType":85,"resourceTypeDesc":86,"sizeBytes":141,"sourceHtmlUrl":142},"executor.py",1116,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fexecutor.py",{"path":144,"resourceType":85,"resourceTypeDesc":86,"sizeBytes":145,"sourceHtmlUrl":146},"judge.py",2677,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fjudge.py",{"path":148,"resourceType":85,"resourceTypeDesc":86,"sizeBytes":149,"sourceHtmlUrl":150},"loop_parallel.py",9466,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Floop_parallel.py",{"path":152,"resourceType":85,"resourceTypeDesc":86,"sizeBytes":153,"sourceHtmlUrl":154},"loop.py",5231,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Floop.py",{"path":156,"resourceType":85,"resourceTypeDesc":86,"sizeBytes":157,"sourceHtmlUrl":158},"multitask_loop.py",15503,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fmultitask_loop.py",{"path":160,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":161,"sourceHtmlUrl":162},"prompts\u002Fbuilder.md",2040,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fprompts\u002Fbuilder.md",{"path":164,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":165,"sourceHtmlUrl":166},"prompts\u002Fexecutor.md",927,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fprompts\u002Fexecutor.md",{"path":168,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":169,"sourceHtmlUrl":170},"prompts\u002Fjudge.md",3702,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fprompts\u002Fjudge.md",{"path":172,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":173,"sourceHtmlUrl":174},"prompts\u002Fproposer.md",1213,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fprompts\u002Fproposer.md",{"path":176,"resourceType":105,"resourceTypeDesc":106,"sizeBytes":177,"sourceHtmlUrl":178},"prompts\u002Fscout.md",797,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fprompts\u002Fscout.md",{"path":180,"resourceType":85,"resourceTypeDesc":86,"sizeBytes":181,"sourceHtmlUrl":182},"proposer.py",1500,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fproposer.py",{"path":184,"resourceType":85,"resourceTypeDesc":86,"sizeBytes":185,"sourceHtmlUrl":186},"workspace.py",3590,"https:\u002F\u002Fgithub.com\u002Fstanfish06\u002Fskillquarium\u002Fblob\u002Fmaster\u002Fskills\u002Fclawpathy-autoresearch\u002Fworkspace.py",[188,195,204,211,218,225],{"skillNo":189,"name":190,"nameZh":6,"description":191,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":192,"categories":193,"keywords":194,"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":196,"name":197,"nameZh":6,"description":198,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":199,"licenseFileUrl":200,"sourceHtmlUrl":201,"categories":202,"keywords":203,"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":205,"name":206,"nameZh":6,"description":207,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":199,"licenseFileUrl":6,"sourceHtmlUrl":208,"categories":209,"keywords":210,"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":212,"name":213,"nameZh":6,"description":214,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":215,"categories":216,"keywords":217,"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":219,"name":220,"nameZh":6,"description":221,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":222,"categories":223,"keywords":224,"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":226,"name":227,"nameZh":6,"description":228,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":199,"licenseFileUrl":229,"sourceHtmlUrl":230,"categories":231,"keywords":232,"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",[],[]]