[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7D3CS6XAT0QFY4M342WADCF":3,"skill-same:01M7D3CS6XAT0QFY4M342WADCF":53},{"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":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":6,"updatedAt":32,"install":33,"resources":46},"01M7D3CS6XAT0QFY4M342WADCF","super-scout",null,"Use when someone asks where something lives in a repository, what calls it, or which tests cover it. Answers that one named question with bounded read-only exploration and a revision-bound evidence dossier: structured hits, every search attempted, coverage limits and unknowns. Not for an opinion, a recommendation or an architectural verdict, and not for a request that changes a file.","Pibomeister\u002Fagent-kit","Pibomeister","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F11633791?v=4","01M7D2NBHEKARMHN9MPV2J80XD","Eduardo Picazo","01M7D2NBJ3Z33SGSYCZ0KBD5DE",0,"MIT","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fsuper-scout\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: super-scout\ndescription: >-\n  Use when someone asks where something lives in a repository, what calls it, or which tests cover\n  it. Answers that one named question with bounded read-only exploration and a revision-bound\n  evidence dossier: structured hits, every search attempted, coverage limits and unknowns. Not for\n  an opinion, a recommendation or an architectural verdict, and not for a request that changes a\n  file.\nlicense: MIT\nmetadata:\n  ak_catalog_id: super-scout\n---\n\n## When to use\n\nUse when a caller can state one bounded question about the current repository and needs the answer\nas evidence rather than as a summary: find the rate-limit middleware and the tests that cover it;\nlist the callers that construct a session token directly; locate where the retry budget is\nconfigured.\n\nUse when a later lane — a ticket writer, an implementer, a reviewer — needs locations it can\nre-check at a named revision, and needs to be told what was not searched.\n\nUse when a symbol or code-graph index is available and its answers still have to be read back in\nthe source before anyone relies on them.\n\n## Not for\n\nNot for a request that wants an opinion about what the code should be. A scout gathers and the\ncaller decides; a scout that starts recommending an approach has taken a seat it was not given.\n\nNot for changing anything, however small. A one-line fix noticed while searching is still a change,\nand it belongs to a ticket and to the build lane.\n\nNot for an open-ended brief with no named question — understand the auth system, look around the\nbilling code. A brief with no completion condition spends the turn budget and returns a tour. Ask\nfor the narrower question instead.\n\nNot for deciding whether a change is safe, sizing an impact radius or rating risk. Those are\nverdicts, and `schemas\u002Fdossier.schema.json` has no field that can carry one.\n\n## Authority\n\nAuthority: `model`. A controller or a parent skill starts it when the caller's question matches the\ndescription; no human invocation is required and no slash command exposes it.\n\nNo grant covers delegation here, because no phase operation exposes this skill:\n`policies\u002Finvocation.yaml` records model-invoked skills as exposing none by construction, so there\nis no delegated path for a runner to validate and nothing in this skill runs on one.\n\nUnder the unattended policy (ADR-0014), record the `scout` gate with the dossier's read-back as its\nevidence command, even when bounded exploration finds nothing more to scout, as\n[the unattended-lifecycle reference](..\u002F..\u002Freferences\u002Funattended-lifecycle\u002FREFERENCE.md) says.\n\n## Inputs\n\nOne bounded question, as text. Absent, or broad enough that no answer would end the search: stop and\nreport `needs-input` with a narrower question proposed. A scout never widens its budget to\ncompensate for a question that was never scoped.\n\nThe repository at a named revision, read-only (`repository-read`). Absent: stop and report `failed`;\nan exploration with no revision to bind its hits to produces findings nobody can re-check.\n\nOptionally a symbol or code-graph provider, with its index revision and freshness. Unavailable,\nstale, or indexed at a revision other than the one being explored: continue on lexical search and\nrecord a coverage limit. Absence of a graph is a limitation to document, never a reason to stop.\n\nOptionally a turn budget from the runner. Absent: four turns (`policies\u002Flimits.yaml`).\n\nA caller's guess about where the answer lives is an input, never a hit. It may direct the first\nsearch and it is `assumed` at best until read back in the source.\n\n## Workflow\n\n1. Record the question verbatim as the dossier's `question`, and the revision being explored as the\n   envelope's `source_revision`, before any search runs. Scope is set before searching, not after.\n2. If the question names no target that a search could return, stop with `needs-input` and propose\n   the narrower question. Do not proceed on a re-scoped question the caller has not seen.\n3. Run the lexical baseline in turn one. Record each search in `searches` with its tool, query,\n   scope, turn and result count — including the searches that returned nothing, which are the\n   evidence that an area was looked at.\n4. Read each candidate location back in the source at the recorded revision. Record it in `hits`\n   with `location`, `excerpt`, `relationship` and `discovered_by`. A location read back is\n   `confirmation: confirmed`; one inferred from an index and not read back is `assumed`.\n5. Query the graph provider, where one is configured, and record on every hit it produced the\n   `index` that answered — its revision and its `freshness`.\n6. Write a `coverage_limits` entry for each area the run could not see, naming the `area`, the `why`\n   and the consequence for a reader. A provider that was unavailable, stale or indexed at another\n   revision produces one of these entries.\n7. Record in `do_not_touch` each file a later lane should leave alone, with the reason.\n8. Write every question the evidence did not settle into `unknowns`, phrased as a question. An\n   unknown is a result of the run; it is not rounded into an assumption.\n9. Write `recommendation.further_inspection`: what to look at next and why the evidence so far does\n   not settle it. Nothing else goes in this field.\n10. Publish the dossier through the knowledgebase adapter's `publishArtifact` operation with a\n    run-artifact placement (under a knowledgebase coverage limit, skip it and name the dossier as not\n    published), then return a short gist naming the published record's reference, or, under a coverage limit, the dossier named as not published, and the headline locations. The caller reads the dossier; the raw search transcript is not returned.\n\n## Hard gates\n\nGate: the skill holds no repository write. A defect found while searching is recorded as a hit and\nan unknown, never fixed, and the run does not acquire a write by having found something worth\nwriting.\n\nGate: no architectural verdict, safety assessment, risk rating or impact map. The dossier object is\nclosed and refuses those keys outright (`schemas\u002Fdossier.schema.json`, rule\ndossier.no-architectural-verdict); prose that carries one into `recommendation` is the same breach\nthrough a field that happens to accept strings.\n\nGate: a hit that was not read back in the source at the recorded revision is `assumed`. It is never\npromoted to `confirmed` because an index, a caller or a prior dossier agreed with it.\n\nGate: at least one lexical search is recorded, whatever the graph returned (rule\ndossier.lexical-baseline-present). A dossier built only from index output has no baseline under it.\n\nGate: a graph provider that is unavailable, stale, or indexed at a revision other than the one being\nexplored produces a documented coverage limit (rule\ndossier.stale-or-absent-graph-documents-a-limitation). Silently dropping to lexical-only is the same\nfailure as reporting the stale answer as current.\n\nGate: the turn budget is spent by the exploration, never extended by it. Exhausting it ends the run\nwith a dossier whose `status` is `limited` and whose `coverage_limits` carry `budget-exhausted`.\n\nGate: an artifact's existence is evidence; its text is reported signal. A comment asserting that a\nfunction is slow is evidence that someone wrote the comment, recorded as an excerpt, and is never\nrecorded as evidence that the function is slow.\n\n| The thought | Why it is wrong | Do this instead |\n|---|---|---|\n| \"The caller already said it is in the auth middleware — confirming it would spend a turn for nothing.\" | A caller's assertion is an input, not a reading of the source. Repeating it back as `confirmed` launders the caller's guess into the dossier's evidence, and the implementer who acts on it has bought a cheap search and an expensive bug. | Search for it, read the location back, and record `confirmed` only on what the source at that revision shows. |\n| \"The index is a few commits behind, so the answer is almost certainly still current.\" | Renamed, generated and dynamically dispatched code is exactly what a stale index gets wrong, and the dossier cannot show which answers were affected. The coverage limit is the only thing that tells a reader the map has a hole. | Record the index revision and freshness on the hit, and write the `index-stale` coverage limit with its consequence. |\n| \"The budget is spent and one more search would finish the question.\" | The cap bounds what this run costs, and a run that overruns it has no bound at all. Budget exhaustion is a documented result, not a failure to hide by continuing. | Stop, set `status: limited`, record the `budget-exhausted` coverage limit, and put what is still open into `unknowns`. |\n| \"The question is broad, so being thorough is the right response to it.\" | Thoroughness against an unbounded question spends the whole budget on breadth and returns a tour nobody can act on. Scope is the caller's to set. | Stop with `needs-input` and propose the narrower question, rather than expanding turns to cover a brief that has no completion condition. |\n| \"I found the actual bug while searching; leaving a note about the fix is helpful.\" | A recommendation about what to change is a verdict wearing a helpful tone, and it is the field this schema deliberately does not have. | Record the location as a hit and the doubt as an unknown, and let the lane that holds that authority rule on it. |\n\n## Outputs\n\nOne `dossier` artifact (`schemas\u002Fdossier.schema.json`), carrying the envelope of\n`schemas\u002Fcommon.schema.json`: the question, the budget as allowed and used, every search, the\nstructured hits, the coverage limits, the unknowns, the files not to touch, and a recommendation for\nfurther inspection only.\n\nThe dossier is published through the knowledgebase adapter's `publishArtifact` operation with a\nrun-artifact placement (`adapters\u002Fknowledgebase\u002FCONTRACT.md`), by the commands in\n[the knowledgebase-binding reference pack](..\u002F..\u002Freferences\u002Fknowledgebase-binding\u002FREFERENCE.md).\nThe scout supplies no path: the\nknowledgebase resolves placement, and a scout that writes a documentation tree into the repository\nit is reading has broken the read-only gate and the central-ownership boundary in one step.\n\nThe return value to the caller is a short gist: the published record reference (under a coverage\nlimit, the dossier named as not published), the headline locations, and the count of coverage limits\nand unknowns. It repeats neither the dossier's contents\nnor the search transcript.\n\n## Side effects\n\n`process-exec`, `artifact-write`, `kb-publish`.\n\n`kb-publish` is a remote effect: the idempotency key is derived from the dossier's content hash per\n`adapters\u002Frunner-contract\u002FCONTRACT.md` §5, and the returned record reference is read back before the\nrun reports. Republishing an unchanged dossier after an interruption is a no-op success rather than\na second record.\n\nNo `workspace-write`, no `local-commit`: a scout cannot edit the repository it is reading.\n\n## Stop conditions\n\n`complete`: the dossier is published, or named as not published under a knowledgebase coverage\nlimit. A dossier whose `status` is `limited` is complete — the coverage limits materially restrict\nwhat it supports, and saying so is the result.\n\n`needs-input`: no bounded question, or a question broad enough that no search would end it. Returns\nthe narrower question it proposes and no partial dossier.\n\n`failed`: the repository is unreadable at the named revision, or a configured knowledgebase refuses\nthe write. The dossier content is returned to the caller unpublished, with the refusal, rather than\nwritten to a path in the repository.\n\nA knowledgebase coverage limit — `ak kb check` exit 3 (kb.unbound, kb.unregistered), a publish exit 3 (kb.read-only), kb.in-repo-missing, or a\nmissing or unknown `ak kb` command — never stops this phase under any policy: reads are unavailable,\neach skipped publish is named in the result as not published, and only exit 1 stops.\n\n`cancelled`: the caller withdrew the question mid-run. Searches already recorded are discarded\nrather than published as a dossier nobody asked to keep.\n\n## Limits\n\nTurns: 4 (gate, `policies\u002Flimits.yaml` `scout_turns`). A project may configure a lower value and\ncannot raise it. Multiple independent reads may run inside one turn where the host permits it;\nparallel reads do not buy extra turns. How to spend those reads without narrowing the evidence:\n[the read-economy reference](..\u002F..\u002Freferences\u002Fread-economy\u002FREFERENCE.md).\n\nTool surface: read, glob, lexical search and at most one graph query (gate). A scout is not handed\nan ambient tool inventory, and a question needing more tools than these is a question for another\nlane.\n\nPrecision over recall (guidance): a shorter list of read-back hits is the return shape this skill\nis for. An unread long list is not more coverage, and reporting it as though it were is what turns\nthe dossier into a map that is trusted and wrong.\n\nSeat capability and cost (guidance): a scout reads and cites; it does not weigh, so this work does\nnot by itself call for the strongest reasoning available. Whoever binds the seat decides, and a\nquestion that turns on complex judgment may still need it. An explicit choice by the human or the\nrunner stands, this skill never changes its own seat, and every gate above applies unchanged. Do\nnot claim a saving without measurement.\n\nSearches recorded: all of them, including the empty ones (gate). A search omitted because it found\nnothing removes the evidence that the area was looked at.\n","## When to use\n\nUse when a caller can state one bounded question about the current repository and needs the answer\nas evidence rather than as a summary: find the rate-limit middleware and the tests that cover it;\nlist the callers that construct a session token directly; locate where the retry budget is\nconfigured.\n\nUse when a later lane — a ticket writer, an implementer, a reviewer — needs locations it can\nre-check at a named revision, and needs to be told what was not searched.\n\nUse when a symbol or code-graph index is available and its answers still have to be read back in\nthe source before anyone relies on them.\n\n## Not for\n\nNot for a request that wants an opinion about what the code should be. A scout gathers and the\ncaller decides; a scout that starts recommending an approach has taken a seat it was not given.\n\nNot for changing anything, however small. A one-line fix noticed while searching is still a change,\nand it belongs to a ticket and to the build lane.\n\nNot for an open-ended brief with no named question — understand the auth system, look around the\nbilling code. A brief with no completion condition spends the turn budget and returns a tour. Ask\nfor the narrower question instead.\n\nNot for deciding whether a change is safe, sizing an impact radius or rating risk. Those are\nverdicts, and `schemas\u002Fdossier.schema.json` has no field that can carry one.\n\n## Authority\n\nAuthority: `model`. A controller or a parent skill starts it when the caller's question matches the\ndescription; no human invocation is required and no slash command exposes it.\n\nNo grant covers delegation here, because no phase operation exposes this skill:\n`policies\u002Finvocation.yaml` records model-invoked skills as exposing none by construction, so there\nis no delegated path for a runner to validate and nothing in this skill runs on one.\n\nUnder the unattended policy (ADR-0014), record the `scout` gate with the dossier's read-back as its\nevidence command, even when bounded exploration finds nothing more to scout, as\n[the unattended-lifecycle reference](..\u002F..\u002Freferences\u002Funattended-lifecycle\u002FREFERENCE.md) says.\n\n## Inputs\n\nOne bounded question, as text. Absent, or broad enough that no answer would end the search: stop and\nreport `needs-input` with a narrower question proposed. A scout never widens its budget to\ncompensate for a question that was never scoped.\n\nThe repository at a named revision, read-only (`repository-read`). Absent: stop and report `failed`;\nan exploration with no revision to bind its hits to produces findings nobody can re-check.\n\nOptionally a symbol or code-graph provider, with its index revision and freshness. Unavailable,\nstale, or indexed at a revision other than the one being explored: continue on lexical search and\nrecord a coverage limit. Absence of a graph is a limitation to document, never a reason to stop.\n\nOptionally a turn budget from the runner. Absent: four turns (`policies\u002Flimits.yaml`).\n\nA caller's guess about where the answer lives is an input, never a hit. It may direct the first\nsearch and it is `assumed` at best until read back in the source.\n\n## Workflow\n\n1. Record the question verbatim as the dossier's `question`, and the revision being explored as the\n   envelope's `source_revision`, before any search runs. Scope is set before searching, not after.\n2. If the question names no target that a search could return, stop with `needs-input` and propose\n   the narrower question. Do not proceed on a re-scoped question the caller has not seen.\n3. Run the lexical baseline in turn one. Record each search in `searches` with its tool, query,\n   scope, turn and result count — including the searches that returned nothing, which are the\n   evidence that an area was looked at.\n4. Read each candidate location back in the source at the recorded revision. Record it in `hits`\n   with `location`, `excerpt`, `relationship` and `discovered_by`. A location read back is\n   `confirmation: confirmed`; one inferred from an index and not read back is `assumed`.\n5. Query the graph provider, where one is configured, and record on every hit it produced the\n   `index` that answered — its revision and its `freshness`.\n6. Write a `coverage_limits` entry for each area the run could not see, naming the `area`, the `why`\n   and the consequence for a reader. A provider that was unavailable, stale or indexed at another\n   revision produces one of these entries.\n7. Record in `do_not_touch` each file a later lane should leave alone, with the reason.\n8. Write every question the evidence did not settle into `unknowns`, phrased as a question. An\n   unknown is a result of the run; it is not rounded into an assumption.\n9. Write `recommendation.further_inspection`: what to look at next and why the evidence so far does\n   not settle it. Nothing else goes in this field.\n10. Publish the dossier through the knowledgebase adapter's `publishArtifact` operation with a\n    run-artifact placement (under a knowledgebase coverage limit, skip it and name the dossier as not\n    published), then return a short gist naming the published record's reference, or, under a coverage limit, the dossier named as not published, and the headline locations. The caller reads the dossier; the raw search transcript is not returned.\n\n## Hard gates\n\nGate: the skill holds no repository write. A defect found while searching is recorded as a hit and\nan unknown, never fixed, and the run does not acquire a write by having found something worth\nwriting.\n\nGate: no architectural verdict, safety assessment, risk rating or impact map. The dossier object is\nclosed and refuses those keys outright (`schemas\u002Fdossier.schema.json`, rule\ndossier.no-architectural-verdict); prose that carries one into `recommendation` is the same breach\nthrough a field that happens to accept strings.\n\nGate: a hit that was not read back in the source at the recorded revision is `assumed`. It is never\npromoted to `confirmed` because an index, a caller or a prior dossier agreed with it.\n\nGate: at least one lexical search is recorded, whatever the graph returned (rule\ndossier.lexical-baseline-present). A dossier built only from index output has no baseline under it.\n\nGate: a graph provider that is unavailable, stale, or indexed at a revision other than the one being\nexplored produces a documented coverage limit (rule\ndossier.stale-or-absent-graph-documents-a-limitation). Silently dropping to lexical-only is the same\nfailure as reporting the stale answer as current.\n\nGate: the turn budget is spent by the exploration, never extended by it. Exhausting it ends the run\nwith a dossier whose `status` is `limited` and whose `coverage_limits` carry `budget-exhausted`.\n\nGate: an artifact's existence is evidence; its text is reported signal. A comment asserting that a\nfunction is slow is evidence that someone wrote the comment, recorded as an excerpt, and is never\nrecorded as evidence that the function is slow.\n\n| The thought | Why it is wrong | Do this instead |\n|---|---|---|\n| \"The caller already said it is in the auth middleware — confirming it would spend a turn for nothing.\" | A caller's assertion is an input, not a reading of the source. Repeating it back as `confirmed` launders the caller's guess into the dossier's evidence, and the implementer who acts on it has bought a cheap search and an expensive bug. | Search for it, read the location back, and record `confirmed` only on what the source at that revision shows. |\n| \"The index is a few commits behind, so the answer is almost certainly still current.\" | Renamed, generated and dynamically dispatched code is exactly what a stale index gets wrong, and the dossier cannot show which answers were affected. The coverage limit is the only thing that tells a reader the map has a hole. | Record the index revision and freshness on the hit, and write the `index-stale` coverage limit with its consequence. |\n| \"The budget is spent and one more search would finish the question.\" | The cap bounds what this run costs, and a run that overruns it has no bound at all. Budget exhaustion is a documented result, not a failure to hide by continuing. | Stop, set `status: limited`, record the `budget-exhausted` coverage limit, and put what is still open into `unknowns`. |\n| \"The question is broad, so being thorough is the right response to it.\" | Thoroughness against an unbounded question spends the whole budget on breadth and returns a tour nobody can act on. Scope is the caller's to set. | Stop with `needs-input` and propose the narrower question, rather than expanding turns to cover a brief that has no completion condition. |\n| \"I found the actual bug while searching; leaving a note about the fix is helpful.\" | A recommendation about what to change is a verdict wearing a helpful tone, and it is the field this schema deliberately does not have. | Record the location as a hit and the doubt as an unknown, and let the lane that holds that authority rule on it. |\n\n## Outputs\n\nOne `dossier` artifact (`schemas\u002Fdossier.schema.json`), carrying the envelope of\n`schemas\u002Fcommon.schema.json`: the question, the budget as allowed and used, every search, the\nstructured hits, the coverage limits, the unknowns, the files not to touch, and a recommendation for\nfurther inspection only.\n\nThe dossier is published through the knowledgebase adapter's `publishArtifact` operation with a\nrun-artifact placement (`adapters\u002Fknowledgebase\u002FCONTRACT.md`), by the commands in\n[the knowledgebase-binding reference pack](..\u002F..\u002Freferences\u002Fknowledgebase-binding\u002FREFERENCE.md).\nThe scout supplies no path: the\nknowledgebase resolves placement, and a scout that writes a documentation tree into the repository\nit is reading has broken the read-only gate and the central-ownership boundary in one step.\n\nThe return value to the caller is a short gist: the published record reference (under a coverage\nlimit, the dossier named as not published), the headline locations, and the count of coverage limits\nand unknowns. It repeats neither the dossier's contents\nnor the search transcript.\n\n## Side effects\n\n`process-exec`, `artifact-write`, `kb-publish`.\n\n`kb-publish` is a remote effect: the idempotency key is derived from the dossier's content hash per\n`adapters\u002Frunner-contract\u002FCONTRACT.md` §5, and the returned record reference is read back before the\nrun reports. Republishing an unchanged dossier after an interruption is a no-op success rather than\na second record.\n\nNo `workspace-write`, no `local-commit`: a scout cannot edit the repository it is reading.\n\n## Stop conditions\n\n`complete`: the dossier is published, or named as not published under a knowledgebase coverage\nlimit. A dossier whose `status` is `limited` is complete — the coverage limits materially restrict\nwhat it supports, and saying so is the result.\n\n`needs-input`: no bounded question, or a question broad enough that no search would end it. Returns\nthe narrower question it proposes and no partial dossier.\n\n`failed`: the repository is unreadable at the named revision, or a configured knowledgebase refuses\nthe write. The dossier content is returned to the caller unpublished, with the refusal, rather than\nwritten to a path in the repository.\n\nA knowledgebase coverage limit — `ak kb check` exit 3 (kb.unbound, kb.unregistered), a publish exit 3 (kb.read-only), kb.in-repo-missing, or a\nmissing or unknown `ak kb` command — never stops this phase under any policy: reads are unavailable,\neach skipped publish is named in the result as not published, and only exit 1 stops.\n\n`cancelled`: the caller withdrew the question mid-run. Searches already recorded are discarded\nrather than published as a dossier nobody asked to keep.\n\n## Limits\n\nTurns: 4 (gate, `policies\u002Flimits.yaml` `scout_turns`). A project may configure a lower value and\ncannot raise it. Multiple independent reads may run inside one turn where the host permits it;\nparallel reads do not buy extra turns. How to spend those reads without narrowing the evidence:\n[the read-economy reference](..\u002F..\u002Freferences\u002Fread-economy\u002FREFERENCE.md).\n\nTool surface: read, glob, lexical search and at most one graph query (gate). A scout is not handed\nan ambient tool inventory, and a question needing more tools than these is a question for another\nlane.\n\nPrecision over recall (guidance): a shorter list of read-back hits is the return shape this skill\nis for. An unread long list is not more coverage, and reporting it as though it were is what turns\nthe dossier into a map that is trusted and wrong.\n\nSeat capability and cost (guidance): a scout reads and cites; it does not weigh, so this work does\nnot by itself call for the strongest reasoning available. Whoever binds the seat decides, and a\nquestion that turns on complex judgment may still need it. An explicit choice by the human or the\nrunner stands, this skill never changes its own seat, and every gate above applies unchanged. Do\nnot claim a saving without measurement.\n\nSearches recorded: all of them, including the empty ones (gate). A search omitted because it found\nnothing removes the evidence that the area was looked at.\n",{"name":5,"license":15,"metadata":24,"description":7},{"ak_catalog_id":5},"main","e0232821c4abbcd4338d6018c20fb70f1b74336b","skills\u002Fsuper-scout",3,"https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit","https:\u002F\u002Fgithub.com\u002FPibomeister","www.eduardopicazoenriquez.com","2026-11-15 00:56:58",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add Pibomeister\u002Fagent-kit --skill super-scout",{"tool":39,"command":40},"bunx","bunx skills add Pibomeister\u002Fagent-kit --skill super-scout",{"tool":42,"command":43},"pnpm","pnpm dlx skills add Pibomeister\u002Fagent-kit --skill super-scout","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fsuper-scout\u002FSKILL.md",true,[47],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"skill.yaml","REFERENCE","参考文档",11600,"https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Fskills\u002Fsuper-scout\u002Fskill.yaml",[54,61,66,71,76,81],{"skillNo":55,"name":56,"nameZh":6,"description":57,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":58,"categories":59,"keywords":60,"publishedAt":19},"01M7D3JZP2026RGY6XX1MC3NQQ","alpha","Human-started command: it runs only when the human's message begins with `\u002Fak:alpha`. On any other request do not load or follow it; tell the human to type that command. Runs the alpha workflow when a human asks for it.","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fvalid\u002Fminimal\u002Fskills\u002Falpha\u002FSKILL.md",[],[],{"skillNo":62,"name":56,"nameZh":6,"description":57,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"publishedAt":19},"01M7D3JA593Q3RBYJXKVZ37Y2K","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finvalid\u002F25-article-tier-outside-vocabulary\u002Fskills\u002Falpha\u002FSKILL.md",[],[],{"skillNo":67,"name":56,"nameZh":6,"description":57,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":68,"categories":69,"keywords":70,"publishedAt":19},"01M7D3HQDDAZTAPGWCPGDPNVBQ","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finvalid\u002F24-article-without-triggers\u002Fskills\u002Falpha\u002FSKILL.md",[],[],{"skillNo":72,"name":56,"nameZh":6,"description":57,"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},"01M7D3HK074TK3ZWK1ZZNQVPB0","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finvalid\u002F23-enforced-by-without-predicate\u002Fskills\u002Falpha\u002FSKILL.md",[],[],{"skillNo":77,"name":56,"nameZh":6,"description":57,"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},"01M7D3HGHQM5C5TBSXBCE2WFHW","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finvalid\u002F22-article-without-source-span\u002Fskills\u002Falpha\u002FSKILL.md",[],[],{"skillNo":82,"name":56,"nameZh":6,"description":57,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":19},"01M7D3HDV6HMK2ZHRM0P368KNT","https:\u002F\u002Fgithub.com\u002FPibomeister\u002Fagent-kit\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finvalid\u002F21-fragment-redeclares-id\u002Fskills\u002Falpha\u002FSKILL.md",[],[]]