[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EF5NA40X2Y4YHRXE5P3MT7":3,"skill-same:01M7EF5NA40X2Y4YHRXE5P3MT7":45},{"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":14,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":29,"ownerDescription":6,"updatedAt":30,"install":31,"resources":44},"01M7EF5NA40X2Y4YHRXE5P3MT7","pharos-investigate",null,"Use when something must be understood before anyone acts: a bug whose cause is not known yet (intermittent, data-dependent, only in one environment, a fix that did not hold), research of data or options for a decision, or understanding how unfamiliar code really works. On \"разберись\", \"почему\", \"найди причину\", \"исследуй\", \"что выбрать\", \"investigate\", \"root cause\".","pharosone\u002Fdev-harness","pharosone","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F286743270?v=4","01M7EF4QCYY7VXGV0CCKEBZT74","PharosOne","01M7EF4QE6P8EG9QQ5B6CA8KWW",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fpharosone\u002Fdev-harness\u002Fblob\u002Fmain\u002Fskills\u002Fpharos-investigate\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: pharos-investigate\ndescription: \"Use when something must be understood before anyone acts: a bug whose cause is not known yet (intermittent, data-dependent, only in one environment, a fix that did not hold), research of data or options for a decision, or understanding how unfamiliar code really works. On \\\"разберись\\\", \\\"почему\\\", \\\"найди причину\\\", \\\"исследуй\\\", \\\"что выбрать\\\", \\\"investigate\\\", \\\"root cause\\\".\"\n---\n\n**Dispatch authority:** resolve every orchestrator\u002Fworker model and effort from\n`harness\u002Fcore\u002Fadapters\u002Fmodels.json` deep-merged with the active\n`harness\u002Fproject\u002Fmodels.json` (or `$HARNESS_PROJECT\u002Fmodels.json`). Record profile, model\nand effort. Never select from personal memory or saved preferences; router shadow is\nadvisory and cannot replace the actual static profile choice. Lane progress goes to\n`\u003Creport>.progress.md`; write final `\u003Creport>` only at the end. Requests use\n`\u003Crun>\u002Flanes\u002F\u003Cwork>.flag.md` or `tripwire.py flag`; never type into another pane.\n\n# pharos-investigate — understand before acting\n\nRead `harness\u002Fcore\u002FREADME.md`, the active runtime adapter and, when present,\n`harness\u002Fproject\u002Fskills\u002Fpharos-investigate.md` (this skill's project specifics) before following\nthis procedure. `harness\u002Fproject` in this skill means the overlay in use: `$HARNESS_PROJECT` when set; `python3 harness\u002Fcore\u002Fscripts\u002Fproject.py overlay` prints it.\n\n**Goal:** an answer someone can act on — the root cause of a bug with its mechanism, the\nnumbers and trade-offs a decision needs, or a true picture of how code behaves — with its\nevidence and an honest confidence. Not a fix, not a spec: those follow from it.\n\n## Efficient testing authority\n\nApply `harness\u002Fcore\u002Frules\u002Fefficient-testing.md` to this workflow's tests and recommendations.\nChoose in-process fakes first, then one shared long-lived test service with owned logical\nisolation, then per-run services, then per-test containers only for a written correctness\nreason. Reuse expensive setup with complete content-hash keys. Keep build\u002Fresult\u002Fdependency\u002F\nimage caches enabled; bypass only a named test\u002Fstep for an input that its cache cannot see.\nRun affected cases\u002Fpackages first locally; the full suite remains required in CI\u002Fpre-PR.\nHeavy gates\u002Fevals\u002Freviews use `harness\u002Fcore\u002Fscripts\u002Fload_guard.py` for adaptive load\u002Fmemory\nadmission, never fixed slots; targeted runs\nand nested admitted work do not wait. Infrastructure changes name the isolation boundary,\ncleanup and measured cold\u002Fwarm wall\u002Fresource cost. Never replace required real paired model\nmeasurements or real-service boundary coverage with a fake.\n\n## Start: the mode question\n\nAsk the depth mode and compare once, with the question tool of the runtime adapter, together\nwith anything else missing (the question, the environment, what may be touched). The\nrecommended option goes first and says why:\n\n- **`standard`** — you investigate yourself on the session's model. Fits when the cause or the\n  answer is likely one or two observations away.\n- **`deep`** — a fresh `investigator` on the `deep-reasoning` profile (Opus `xhigh` \u002F Astra\n  `xhigh`) does one deep pass, and the other vendor gives a second opinion on its report before\n  it reaches the user. Recommend it when a wrong answer is expensive or the obvious answer\n  already failed: the cause is not established after a first look; the failure is intermittent,\n  concurrent, data- or environment-dependent; a previous fix did not hold; the decision is\n  costly to reverse or needs data analysed; the code spans unfamiliar layers.\n\nWrite the recommendation as \"`deep` (Recommended) — \u003Cthe signal you saw>\" or the same for\n`standard`.\n\nAsk `compare` in the same question; the default is `off`:\n\n- **`off`** (default) — the investigation as described here.\n- **`investigation`** — two independent investigators, one per vendor, answer the same frozen\n  brief without seeing each other's report; then each vendor critiques the other's report, and\n  the two critiques become a disagreement map. Recommend it when the decision is costly to\n  reverse, there is conflicting evidence, or a deep pass already left low confidence:\n  \"`compare=investigation` (Recommended) — \u003Cthe signal>\". Without that signal recommend `off`.\n\n`compare=off` leaves the flow unchanged: `deep` is one investigator and a second opinion, as\nbefore. Record the answers, model and effort in `run.md`\n(`python3 harness\u002Fcore\u002Fscripts\u002Ftask_state.py init \u003Ctask>`) and as the run's first event:\n`task_state.py event \u003Ctask> run.start --data '{\"workflow\": \"investigate\", \"depth\": \"\u003Cmode>\",\n\"compare\": \"off|investigation\", \"runtime_mode\": \"\u003Cmode>\"}'`.\n\n## Run it\n\n1. **Frame:** the question in one sentence and what observation settles it (the investigator\n   role defines the frame, the hypotheses and the report; follow it in `standard` mode too).\n2. **Evidence:** reproduce, measure, query, bisect, instrument — in a scratch directory or a\n   worktree (`python3 harness\u002Fcore\u002Fscripts\u002Fworktree.py add \u003Cname>`), never by editing the\n   checkout under investigation. Production data and systems only read-only and only as the\n   user allowed; no secrets in briefs, reports or logs; never deploy.\n3. **`deep`:** brief the `investigator` with the question, what is known with sources, the\n   environment and its limits, and the report path `\u003Crun-dir>\u002Finvestigation.md`. Then the\n   second opinion: the other vendor's `investigator` gets the question and the report and is\n   asked to break its conclusion — evidence that does not hold, a hypothesis not ruled out, a\n   cheaper explanation. Claude-only or Codex-only mode: a fresh context of the same vendor,\n   recorded as single-vendor. Reconcile: a challenge that stands goes back to the investigator\n   once, or into the report's Unknowns.\n   **`compare=investigation`** replaces step 3. Write the brief to `\u003Crun-dir>\u002Fbrief.md`: a\n   `## Question` section, what is known with sources, the environment and its limits; it is\n   frozen (hashed, given unchanged to both). Run\n   `python3 harness\u002Fcore\u002Fscripts\u002Fcompare.py investigation --task \u003Ctask> --input\n   \u003Crun-dir>\u002Fbrief.md --checkout .`; exit 2 is bad input and nothing was launched; a stopped\n   compare continues with `--resume \u003Cid>`. It writes two reports, a critique of each by the\n   other vendor, and in `\u003Crun-dir>\u002Fcompare\u002F\u003Cid>\u002F` the disagreement map (`disagreements.md`:\n   claims agreed, claims in dispute with both sides' evidence, the cheapest observation that\n   settles each) and `comparison.md` (both critiques).\n   Show the user the map and the critiques verbatim. A disputed answer means low confidence\n   until the settling observation is made; a missing report or critique is said aloud.\n4. **Report to the user:** the answer, the confidence and what would change it, the evidence in\n   short, the recommendation and the open unknowns (with compare: the disputed claims and their\n   settling observations). A confidence below high says what it would cost to raise it.\n\n## Hand-off\n\n- A bug: the reproduction becomes the failing test of a `pharos-dev` fix unit.\n- A decision: the options with their numbers, recommended first, as a question to the user;\n  the answer is recorded before dependent work.\n- A feature or redesign: `pharos-discover`, with the report as the design's evidence.\n\n## Stop and ask\n\n- The evidence needs access the user has not granted (production, paid compute, private data).\n- Two passes left the answer at low confidence: present what is known and the cost of the next\n  step instead of a third pass.\n","**Dispatch authority:** resolve every orchestrator\u002Fworker model and effort from\n`harness\u002Fcore\u002Fadapters\u002Fmodels.json` deep-merged with the active\n`harness\u002Fproject\u002Fmodels.json` (or `$HARNESS_PROJECT\u002Fmodels.json`). Record profile, model\nand effort. Never select from personal memory or saved preferences; router shadow is\nadvisory and cannot replace the actual static profile choice. Lane progress goes to\n`\u003Creport>.progress.md`; write final `\u003Creport>` only at the end. Requests use\n`\u003Crun>\u002Flanes\u002F\u003Cwork>.flag.md` or `tripwire.py flag`; never type into another pane.\n\n# pharos-investigate — understand before acting\n\nRead `harness\u002Fcore\u002FREADME.md`, the active runtime adapter and, when present,\n`harness\u002Fproject\u002Fskills\u002Fpharos-investigate.md` (this skill's project specifics) before following\nthis procedure. `harness\u002Fproject` in this skill means the overlay in use: `$HARNESS_PROJECT` when set; `python3 harness\u002Fcore\u002Fscripts\u002Fproject.py overlay` prints it.\n\n**Goal:** an answer someone can act on — the root cause of a bug with its mechanism, the\nnumbers and trade-offs a decision needs, or a true picture of how code behaves — with its\nevidence and an honest confidence. Not a fix, not a spec: those follow from it.\n\n## Efficient testing authority\n\nApply `harness\u002Fcore\u002Frules\u002Fefficient-testing.md` to this workflow's tests and recommendations.\nChoose in-process fakes first, then one shared long-lived test service with owned logical\nisolation, then per-run services, then per-test containers only for a written correctness\nreason. Reuse expensive setup with complete content-hash keys. Keep build\u002Fresult\u002Fdependency\u002F\nimage caches enabled; bypass only a named test\u002Fstep for an input that its cache cannot see.\nRun affected cases\u002Fpackages first locally; the full suite remains required in CI\u002Fpre-PR.\nHeavy gates\u002Fevals\u002Freviews use `harness\u002Fcore\u002Fscripts\u002Fload_guard.py` for adaptive load\u002Fmemory\nadmission, never fixed slots; targeted runs\nand nested admitted work do not wait. Infrastructure changes name the isolation boundary,\ncleanup and measured cold\u002Fwarm wall\u002Fresource cost. Never replace required real paired model\nmeasurements or real-service boundary coverage with a fake.\n\n## Start: the mode question\n\nAsk the depth mode and compare once, with the question tool of the runtime adapter, together\nwith anything else missing (the question, the environment, what may be touched). The\nrecommended option goes first and says why:\n\n- **`standard`** — you investigate yourself on the session's model. Fits when the cause or the\n  answer is likely one or two observations away.\n- **`deep`** — a fresh `investigator` on the `deep-reasoning` profile (Opus `xhigh` \u002F Astra\n  `xhigh`) does one deep pass, and the other vendor gives a second opinion on its report before\n  it reaches the user. Recommend it when a wrong answer is expensive or the obvious answer\n  already failed: the cause is not established after a first look; the failure is intermittent,\n  concurrent, data- or environment-dependent; a previous fix did not hold; the decision is\n  costly to reverse or needs data analysed; the code spans unfamiliar layers.\n\nWrite the recommendation as \"`deep` (Recommended) — \u003Cthe signal you saw>\" or the same for\n`standard`.\n\nAsk `compare` in the same question; the default is `off`:\n\n- **`off`** (default) — the investigation as described here.\n- **`investigation`** — two independent investigators, one per vendor, answer the same frozen\n  brief without seeing each other's report; then each vendor critiques the other's report, and\n  the two critiques become a disagreement map. Recommend it when the decision is costly to\n  reverse, there is conflicting evidence, or a deep pass already left low confidence:\n  \"`compare=investigation` (Recommended) — \u003Cthe signal>\". Without that signal recommend `off`.\n\n`compare=off` leaves the flow unchanged: `deep` is one investigator and a second opinion, as\nbefore. Record the answers, model and effort in `run.md`\n(`python3 harness\u002Fcore\u002Fscripts\u002Ftask_state.py init \u003Ctask>`) and as the run's first event:\n`task_state.py event \u003Ctask> run.start --data '{\"workflow\": \"investigate\", \"depth\": \"\u003Cmode>\",\n\"compare\": \"off|investigation\", \"runtime_mode\": \"\u003Cmode>\"}'`.\n\n## Run it\n\n1. **Frame:** the question in one sentence and what observation settles it (the investigator\n   role defines the frame, the hypotheses and the report; follow it in `standard` mode too).\n2. **Evidence:** reproduce, measure, query, bisect, instrument — in a scratch directory or a\n   worktree (`python3 harness\u002Fcore\u002Fscripts\u002Fworktree.py add \u003Cname>`), never by editing the\n   checkout under investigation. Production data and systems only read-only and only as the\n   user allowed; no secrets in briefs, reports or logs; never deploy.\n3. **`deep`:** brief the `investigator` with the question, what is known with sources, the\n   environment and its limits, and the report path `\u003Crun-dir>\u002Finvestigation.md`. Then the\n   second opinion: the other vendor's `investigator` gets the question and the report and is\n   asked to break its conclusion — evidence that does not hold, a hypothesis not ruled out, a\n   cheaper explanation. Claude-only or Codex-only mode: a fresh context of the same vendor,\n   recorded as single-vendor. Reconcile: a challenge that stands goes back to the investigator\n   once, or into the report's Unknowns.\n   **`compare=investigation`** replaces step 3. Write the brief to `\u003Crun-dir>\u002Fbrief.md`: a\n   `## Question` section, what is known with sources, the environment and its limits; it is\n   frozen (hashed, given unchanged to both). Run\n   `python3 harness\u002Fcore\u002Fscripts\u002Fcompare.py investigation --task \u003Ctask> --input\n   \u003Crun-dir>\u002Fbrief.md --checkout .`; exit 2 is bad input and nothing was launched; a stopped\n   compare continues with `--resume \u003Cid>`. It writes two reports, a critique of each by the\n   other vendor, and in `\u003Crun-dir>\u002Fcompare\u002F\u003Cid>\u002F` the disagreement map (`disagreements.md`:\n   claims agreed, claims in dispute with both sides' evidence, the cheapest observation that\n   settles each) and `comparison.md` (both critiques).\n   Show the user the map and the critiques verbatim. A disputed answer means low confidence\n   until the settling observation is made; a missing report or critique is said aloud.\n4. **Report to the user:** the answer, the confidence and what would change it, the evidence in\n   short, the recommendation and the open unknowns (with compare: the disputed claims and their\n   settling observations). A confidence below high says what it would cost to raise it.\n\n## Hand-off\n\n- A bug: the reproduction becomes the failing test of a `pharos-dev` fix unit.\n- A decision: the options with their numbers, recommended first, as a question to the user;\n  the answer is recorded before dependent work.\n- A feature or redesign: `pharos-discover`, with the report as the design's evidence.\n\n## Stop and ask\n\n- The evidence needs access the user has not granted (production, paid compute, private data).\n- Two passes left the answer at low confidence: present what is known and the cost of the next\n  step instead of a third pass.\n",{"name":5,"description":7},"main","a88770aee5bc4b664c660ce05b4da13854e51726","skills\u002Fpharos-investigate","https:\u002F\u002Fgithub.com\u002Fpharosone\u002Fdev-harness","https:\u002F\u002Fgithub.com\u002Fpharosone","https:\u002F\u002Fpharosone.ai\u002F","2026-10-14 19:18:56",{"source":8,"skillName":5,"commands":32,"prompt":42,"zipAvailable":43},[33,36,39],{"tool":34,"command":35},"npx","npx skills add pharosone\u002Fdev-harness --skill pharos-investigate",{"tool":37,"command":38},"bunx","bunx skills add pharosone\u002Fdev-harness --skill pharos-investigate",{"tool":40,"command":41},"pnpm","pnpm dlx skills add pharosone\u002Fdev-harness --skill pharos-investigate","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fpharosone\u002Fdev-harness\u002Fblob\u002Fmain\u002Fskills\u002Fpharos-investigate\u002FSKILL.md",true,[],[46,53,60,67,74,81],{"skillNo":47,"name":48,"nameZh":6,"description":49,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":50,"categories":51,"keywords":52,"publishedAt":19},"01M7EF616KZ861AS6GKEWX1B8W","deploy-app","Use when the fixture app has to be deployed to its staging host.","https:\u002F\u002Fgithub.com\u002Fpharosone\u002Fdev-harness\u002Fblob\u002Fmain\u002Ftests\u002Ffixture\u002Fharness\u002Fproject\u002Fskills\u002Fdeploy-app\u002FSKILL.md",[],[],{"skillNo":54,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":19},"01M7EF5VNNSHBGNMWW5M9RR0V9","submit-pr","Use when the user asks to open, update or finalize a pull request for work in this repository (\"сделай PR\", \"открой PR\", \"submit\", \"push and open a PR\"). Never on its own initiative.","https:\u002F\u002Fgithub.com\u002Fpharosone\u002Fdev-harness\u002Fblob\u002Fmain\u002Fskills\u002Fsubmit-pr\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":63,"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},"01M7EF5EN0FX3R1Y4AHFFBM9J1","pharos-discover","Use when a feature, endpoint, screen or behaviour change in this repository needs requirements, alternative approaches and an approved specification with TDD units; also when a spec lacks its Requirements interview, its Placement, its Taste gates or a section a project interview row requires.","https:\u002F\u002Fgithub.com\u002Fpharosone\u002Fdev-harness\u002Fblob\u002Fmain\u002Fskills\u002Fpharos-discover\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":19},"01M7EF5729KFDQ8CGB6QS5C3E4","pharos-dev","Use when implementing, fixing or refactoring anything in this repository beyond a one-line edit — executing an approved spec, a feature, a bugfix — and on follow-ups (\"сделай\", \"реализуй\", \"исполни спеку\", \"почини\", \"продолжи\", \"перезапусти ревью\", \"redo the QA\"). A review-only request runs the review step alone.","https:\u002F\u002Fgithub.com\u002Fpharosone\u002Fdev-harness\u002Fblob\u002Fmain\u002Fskills\u002Fpharos-dev\u002FSKILL.md",[],[],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":19},"01M7EF531SZB34M9RYAJ28J056","manual-test-plan","Use when a PR in this repository needs its hand-executable verification plan written or executed (\"test plan\", \"план ручного тестирования\", \"lets start test plan\", \"начнём тест-план\"); submit-pr invokes it for every PR.","https:\u002F\u002Fgithub.com\u002Fpharosone\u002Fdev-harness\u002Fblob\u002Fmain\u002Fskills\u002Fmanual-test-plan\u002FSKILL.md",[],[],{"skillNo":82,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":19},"01M7EF4WYM7SJ67WY1306WKEH9","design-review","Use when a PR, branch or diff in this repository needs an architecture pass — leaking abstractions, a layer owning what another should, what became public, drift between sibling implementations. On \"design review\", \"architecture review\", \"ревью дизайна\", \"архитектурное ревью\", \"проверь абстракции\".","https:\u002F\u002Fgithub.com\u002Fpharosone\u002Fdev-harness\u002Fblob\u002Fmain\u002Fskills\u002Fdesign-review\u002FSKILL.md",[],[]]