[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6CHCGW422AP45MEGSQ37THH":3,"skill-same:01M6CHCGW422AP45MEGSQ37THH":94},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":5,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":22,"publishedAt":42,"linkOnly":43,"markdownContent":44,"markdownBody":45,"frontMatter":46,"refName":48,"commitSha":49,"skillDirectory":50,"forksCount":51,"repositoryHtmlUrl":52,"repositoryHomepageUrl":53,"ownerHtmlUrl":54,"ownerBlogUrl":53,"ownerDescription":55,"updatedAt":56,"install":57,"resources":70},"01M6CHCGW422AP45MEGSQ37THH","skilld",null,"Operate skilld CLI for Skill discovery, use, installation, inspection, updates, authentication, configuration, restoration, and removal, including Repository, curator, and collection refs. Also read the skilld.dev registry (view, browse, trending, tracks, curators, index) and act for the user's skilld.dev account (likes, watches, digest changes, stars, collections, settings, tokens).","skilld-dev\u002Fskilld","skilld-dev","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F272400311?v=4","01M6CHC10A2J387ZK0JM77EDFJ","01M6CHC14CQ0B2KY6VFD07YDHE",332,"MIT","https:\u002F\u002Fgithub.com\u002Fskilld-dev\u002Fskilld\u002Fblob\u002Fmain\u002Fskills\u002Fskilld\u002FSKILL.md",[17],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[5,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41],"browse","configuration","skill discovery","use","settings","view","read","filesystem","access","cli","shell","discovery","collection","network","repository","index","dev","registry","authentication","2026-11-10 10:12:30",false,"---\nname: skilld\ndescription: Operate skilld CLI for Skill discovery, use, installation, inspection, updates, authentication, configuration, restoration, and removal, including Repository, curator, and collection refs. Also read the skilld.dev registry (view, browse, trending, tracks, curators, index) and act for the user's skilld.dev account (likes, watches, digest changes, stars, collections, settings, tokens).\nlicense: MIT\ncompatibility: \"Requires an agent with filesystem and shell access. Network access is needed for remote sources.\"\n---\n\n# Use skilld CLI\n\nUse skilld CLI to find, load, install, inspect, restore, update, verify, and remove Skills.\n\nRun a Skill first. Install a Skill only when the user asks to keep it.\nIf the user asks to fork a Skill, copy its source before installing the local copy.\n\n## Use Agent output\n\nUse `--json` with `search`, `run`, and `update --check`.\nUse `--json` with `view` of a registry ref, and with every registry and account command below.\nTheir `data` is the skilld.dev answer. A command whose answer has no body returns `data: null`.\nUse `--json` with `sync` and `sync --check` for declared Skills.\nUse `doctor --json` for read-only Skill discovery and cleanup findings.\nThe remaining commands do not support JSON output.\nUse `--plain` when another command needs stable text.\n\nCheck the exit code before reading stdout.\nRead JSON success data only when `_tag` is `Success`.\nRead JSON failures from stderr.\nReport the error `code` and `message`.\nA failed `skilld run` ends with a `Next step:` line. In JSON, read `error.nextStep`.\nFollow that step. It says whether to retry, when, and what to tell the user.\nExit code `75` means a temporary failure, such as a network fault or a busy service.\nExit code `1` means the same command gives the same result.\nIf `skilld run` names files it loaded without, tell the user when the instructions need one.\n\nAn update check can exit with code 1 and return valid JSON.\nRead its update relations before treating that exit as a failure.\nNever parse formatted terminal output.\n\n## Sync declared Skills\n\nRead [Declared Skills](references\u002Fdeclared-skills.md) when a project needs explicit Skill requirements.\nUse `.skills\u002Fskilld.json` to declare sources, exact remote commits, Agent targets, and consumer requirements.\nRun `skilld sync --check --json` to find differences without installing or fetching remote bytes.\nExit code `1` with success data means sync is needed.\nRun `skilld sync` to install the full declaration in one transaction.\nUse `--global` for global Agent targets.\nUse `--adopt` only when migrating identical unmanaged symlinks.\nChanged unmanaged targets block sync.\nIf the project declares this Skill, use its declaration to refresh it.\nDo not install a second copy through another store.\nRequired Skills cannot be removed until their declaration releases the requirement.\n\nFor browser-controlled account login, use `skilld auth login --no-browser --plain`.\nOpen the printed URL in the intended signed-in browser while the command waits.\n\n## Search for a Skill\n\nRun a focused search:\n\n```sh\nskilld search \u003Cquery> --json\n```\n\nRead `data.items` before choosing a Skill.\nUse each item's `selector` for a Skill run.\nRefine the query when several Skills cover different tasks.\n\nDo not guess a selector from the Skill name.\nDo not install a search result before reading its description.\n\n## Run a Skill\n\nRun the selector returned by search:\n\n```sh\nskilld run \u003Cselector> --json\n```\n\nThe command prints SKILL.md and writes no Skill files.\nIt retains no remote Skill files after the command exits.\nRead the printed SKILL.md, then follow it for the current task.\nPrefer `skilld run` for a one-off task.\n\nRead `data.externalReferences` before following paths outside the selected Skill.\nThese notices identify references, not verified dependencies.\nFor a possible remote sibling Skill, `readArgv` pins its Repository and commit.\nRun that array only when the task and access permissions allow the read.\nNever add `--direct` to bypass an access failure.\nFor local references, resolve from the original Skill directory, not an Agent target copy.\nIf a reference is unresolved, identify its source before following it.\nInstall extra Skills only when the user asks to keep them.\n\nRead `data.files` for each supporting file's path, kind, and size.\nThe initial load prints no supporting file content.\nRead one only when the instructions name it:\n\n```sh\nskilld run \u003Cselector> --revision \u003Cdata.revision> --file \u003Cpath> --json\n```\n\nRun the exact `data.files[].readArgv` array when possible.\nIt contains the source, exact revision, file path, and `--json`.\nRepeat `--file` to read several files in one command.\n\nCheck `data.files[].readable` before you ask for a file.\nA file with `readable: false` never prints.\nIts `kind` is `executable` or `binary`.\nTell the user the Skill needs an install to use that file.\n\nRead `data.behaviors` before you follow the Skill.\nEach entry names one behavior, its `tier`, and the `path` and `line` where it appears.\nTell the user about every behavior before you act on it.\nAn empty list proves nothing. Patterns miss obfuscated code.\n\nA remote run stops with `BEHAVIOR_CONFIRMATION_REQUIRED` when the Skill has an `ask` behavior.\nskilld loaded nothing. Show the user every behavior in `error.message`.\nA match can carry a model reading, such as `(model reading: quoted example. REASON)`.\nShow it as a language model's reading. It never replaces the user's approval.\nIf the user approves, run the command at the end of the message and add `--json`.\nNever add `--allow` to any command without the user's approval in this session.\nIf the user declines, stop and load nothing.\n\nReport which Skill you ran and that skilld wrote no Skill files.\nRead `data.sourceStatus`, `data.origin`, and `data.revision`.\nA `verified` status covers where the Skill came from.\nIt does not cover what the instructions ask you to do.\nIf the status is `unverified`, tell the user before you follow the Skill.\n\n## Author a Skill from the user's own code\n\nThe skilld CLI has no generation command.\nAuthoring lives in skilld-maintained Skills that you run.\n\nRun the Skill that matches the request:\n\n```sh\nskilld run skilld-dev\u002Fskilld\u002Fgenerate-project-skill --json\nskilld run skilld-dev\u002Fskilld\u002Fgenerate-package-skill --json\nskilld run skilld-dev\u002Fskilld\u002Freview-skill --json\n```\n\nUse `generate-project-skill` for the current project or workspace.\nUse `generate-package-skill` for a package the user maintains.\nUse `review-skill` before the user publishes a Skill.\n\nRead the printed SKILL.md, then follow it for this project.\nWrite the draft where the user asks, and show the files for review.\nDo not replace an existing Skill until the user approves the draft.\nDo not install or publish a draft the user has not read.\n\nThe user owns the draft. They commit it to their own Repository.\n\n## List the Skills a Repository, curator, or collection names\n\nskilld.dev prints refs that name several Skills:\n\n- `OWNER\u002FREPOSITORY` names every Skill in one Repository.\n- `@LOGIN` names every Skill in one curator's collections.\n- `@LOGIN\u002FSLUG` names every Skill in one collection.\n\nRun one of these refs to list its Skills:\n\n```sh\nskilld run @LOGIN\u002FSLUG --json\n```\n\nThe command prints an index and loads no Skill.\nRead `data.items` for each Skill's `name`, `owner`, `repository`, `description`, and `selector`.\nRun the `data.items[].runArgv` array to load one Skill.\nPick the Skills the current task needs. Do not run every Skill in the index.\n\n## Read the registry\n\nRead one registry entry before you run or install it:\n\n```sh\nskilld view OWNER\u002FREPOSITORY\u002FSKILL --json\n```\n\nRead `data.owner`, `data.sourceUrl`, `data.sourceCommit`, and `data.description`.\nReport the author and the exact SKILL.md with the Skill.\n`skilld view` also takes `OWNER\u002FREPOSITORY`, `@LOGIN`, and `@LOGIN\u002FSLUG`.\nA bare name without `\u002F` or `@` shows an installed Skill.\n\nUse these commands when the user asks what exists, not for one task:\n\n```sh\nskilld browse \u003Cquery> --sort stars --json\nskilld trending --json\nskilld tracks --json\nskilld tracks \u003Cslug> --json\nskilld curators --json\n```\n\nUse `skilld search` to find a Skill for the current task.\nEach trending item has a `signal`. Its `kind` says why the Skill trends.\nReport that reason with the Skill.\nUse `--limit` and `--offset` to page. `data.total` counts every result.\n\nThe user can name a Repository the registry does not list. Ask skilld.dev to index it:\n\n```sh\nskilld index OWNER\u002FREPOSITORY --json\n```\n\nThe command waits about a minute.\nIf `data.status` is still `queued`, run the same command again later.\nAn `INDEX_FAILED` error names the reason.\n\n## Choose the source\n\nPrefer the exact `OWNER\u002FREPOSITORY\u002FSKILL` selector returned by Skill search.\nHosted selectors use immutable artifact delivery from an exact Git commit.\n\nUse a local path only for a Skill the user already controls:\n\n```sh\nskilld run .\u002Fskills\u002Fmy-skill --json\nskilld install .\u002Fskills\u002Fmy-skill\n```\n\nUse `--direct` only for an explicit public GitHub selector.\nDirect mode bypasses artifact delivery and gives the `unverified` source status.\nIt cannot access a private repository.\nNever add `--direct` merely to bypass a delivery failure.\n\nPrivate repository delivery requires a skilld.dev account and GitHub App access.\nIf private access fails, check authentication before changing the selector.\n\n## Install a Skill\n\nInstall a Skill when the user wants it in every session.\nInstall a Skill when it must run its own script.\nAn explicit install or fork request authorizes the requested project files.\nOtherwise, ask before installing.\n\nInstall the selector returned by search into the detected Agent target:\n\n```sh\nskilld install \u003Cselector>\n```\n\nThe default scope is the current project.\nProject installs update `.skills\u002Fskilld-lock.yaml` and selected Agent targets.\n\nInstall into global Agent targets:\n\n```sh\nskilld install \u003Cselector> --global\n```\n\nUse `--agent \u003Cagent>` when the user names an Agent target.\nRepeat `--agent` when the user names several Agent targets.\nDo not guess a target when detection and `agent.targets` are empty.\n\nUse `--mode copy` or `--mode symlink` only when the user chooses a mode.\nOtherwise, use the configured `install.mode`.\n\nAn install stops with `BEHAVIOR_CONFIRMATION_REQUIRED` when the Skill has an `ask` behavior.\nskilld wrote nothing. Show the user every behavior in the message.\nIf the user approves, run the same command again with the `--allow` ids the message names.\n`skilld add` installs the other Skills and lists each held Skill at the end.\n\nInstall this skilld-maintained Skill globally:\n\n```sh\nskilld install skilld --global\n```\n\nInstall every Skill a Repository, curator, or collection names:\n\n```sh\nskilld add OWNER\u002FREPOSITORY --all\nskilld add @LOGIN\u002FSLUG --all --global\n```\n\n`skilld add` accepts `--global`, `--agent`, and `--mode` like `skilld install`.\nIt prints one `Installed Skill` line per Skill.\nIf several Skills are listed, an Agent run requires `--all`.\nA normal terminal asks which Skills to install. `--plain` also requires `--all` for several Skills.\nOne listed Skill installs without a picker. `add` does not support JSON output.\nRun `skilld run` with the same ref first, then confirm the list with the user before using `--all`.\n`skilld add` with one Skill selector installs that Skill like `skilld install`.\nListing never requests registry indexing.\nSkills discovered through GitHub still use hosted delivery by default.\nIf delivery fails, skilld shows an explicit direct installation command where possible.\nOnly use direct installation when the user chooses it. Never use it to bypass a verification or policy failure.\n`skilld add OWNER\u002FREPOSITORY --all --direct` lists and installs public GitHub files without skilld.dev.\nIt records `unverified`. Curator and collection refs require hosted delivery.\nBehavior approval still applies.\n`skilld add .\u002FPATH` installs one local Skill whose directory contains `SKILL.md`.\n\nAlways use the source selector shown by `skilld search`.\nAfter installation, report the Skill name, scope, installed agent paths, and source status.\n\n## Fork a Skill\n\nTreat `fork this Skill \u003Cskilld.dev URL>` as a request for an editable local Skill.\nCreate no GitHub fork unless the user asks for one.\nCopy the Skill before following any of its instructions.\n\n1. Read the Skill page with `Accept: text\u002Fmarkdown`, or fetch its `.md` URL.\n   For `\u002Fgh\u002FOWNER\u002FREPOSITORY\u002FSKILL`, use `skilld view OWNER\u002FREPOSITORY\u002FSKILL --json` for source metadata.\n   Check the exit code and `_tag` before reading `data`.\n2. Read `sourceUrl`, `sourceCommit`, `skillPath`, `sourceGone`, and `license`.\n   Use the repository identified by `sourceUrl`, including any GitHub rename.\n   If the source is gone or its path is missing, stop and report the failure.\n   If `license` is null, read licence files at the source commit.\n3. Fetch the source into a temporary directory using Git.\n   Check out `sourceCommit` detached.\n   If it is absent, resolve the ref in `sourceUrl` to one commit before copying.\n   Record that actual commit. Never combine files from different commits.\n   Read the source licence files before creating the local copy.\n   If copying is not permitted, stop and report the restriction.\n4. Copy the directory containing `skillPath` into `.\u002Fskills\u002FSKILL`, unless the user chose another path.\n   Copy the original SKILL.md and all supporting files, including scripts and binary assets.\n   Preserve executable modes. Exclude `.git` metadata.\n   Do not save the page wrapper or its rewritten links as SKILL.md.\n   Reject symlinks and paths outside the Skill directory.\n   Never overwrite an existing directory or Agent target.\n5. Preserve the original author, licence, and notices.\n   Include applicable licence files from the repository or parent directories.\n   Add `PROVENANCE.md` with the Skill page, source URL, actual commit, original path, and licence.\n   If that file already exists, retain it and record provenance in a separate file.\n   Do not replace the original author's credit with the user's name.\n6. In the project root, run `skilld install .\u002Fskills\u002FSKILL --mode copy`.\n   First inspect the project lockfile and selected Agent target directories for this Skill name.\n   If it is already installed, stop before replacing it.\n   Use the detected Agent targets. Pass `--agent` only for targets the user selected.\n   This records a local source. Never install the upstream selector for a fork.\n   If target selection or installation fails, preserve the local copy and report the exact failure.\n7. Report the local path, source commit, and installed Agent targets.\n   After local edits, reinstall the same local path to refresh Agent targets.\n   Upstream updates must not replace the local copy.\n\nDo not publish or push the local copy unless the user asks.\n\n## Restore locked Skills\n\nRestore the current project from its lockfile:\n\n```sh\nskilld install\n```\n\nRestore global Skills from the global scope:\n\n```sh\nskilld install --global\n```\n\nA verified remote Skill restores its exact locked commit through artifact delivery.\nAn unverified remote Skill requires the recovery command shown by skilld.\nDo not convert a verified source to direct mode during recovery.\n\nNever delete a lockfile or Agent target to repair an install.\nPreserve the files and report the exact failure first.\n\n## Inspect installed Skills\n\n```sh\nskilld list\nskilld list --global\nskilld view \u003Cskill>\nskilld view \u003Cskill> --global\n```\n\nUse `list` to find installed names in one scope.\nUse `view` to inspect a Skill before any mutation.\nRead its path, source, source status, and Agent targets.\n\n## Check and apply updates\n\n```sh\nskilld update --check --json\nskilld update \u003Cskill>\nskilld update \u003Cskill> --global\n```\n\nUse `update --check --json` to inspect update relations without changing files.\nRead each `data.items[].relation._tag` before changing files.\nUse `update \u003Cskill>` only when the relation is `available`.\nTreat `current`, `pinned`, and `notTracked` as no action.\nIf the relation is `behind` or `diverged`, ask before changing files.\nIf the relation is `unavailable`, report `failure.code` and `failure.message`.\nTreat `unavailable` as unknown. Do not infer a newer commit.\n\nAn update stops with `BEHAVIOR_CONFIRMATION_REQUIRED` when the new version adds an `ask` behavior.\nskilld changed nothing. Show the user every behavior in the message.\nIf the user approves, run the same command again with the `--allow` ids the message names.\n\nUpdate one named Skill unless the user explicitly requests all updates.\nUse `--global` only for a Skill in the global scope.\nLeave `--interactive` to a human terminal session.\n\n## Verify source integrity\n\n```sh\nskilld verify \u003Cskill>\n```\n\nUse `verify` to check installed bytes against recorded source data.\nA successful check confirms provenance and integrity only.\nIt does not approve the Skill instructions.\n\nIf verification fails, do not hand edit a managed Skill.\nUse `view` to inspect its source before update or restore.\n\n## Report outdated and unmanaged Skills\n\nCheck the current scope:\n\n```sh\nskilld outdated --plain\n```\n\nCheck both scopes and every Agent target directory:\n\n```sh\nskilld outdated --all --plain\n```\n\nUse `outdated` for stale, unverified, local, and unmanaged Skill reports.\nRead every proposed command before using it.\nNever delete an unmanaged Skill unless the user names it for removal.\n\n## Review Skill files across projects\n\nRun `skilld doctor --json` to scan the user's home directory.\nPass explicit roots to narrow the scan or inspect files outside home.\nThe default scan excludes dependencies, caches, generated files, backups, and Git worktrees.\nUse `--exclude GLOB` to prune more paths.\nUse `--include-excluded` or `--include-worktrees` only when those files matter to the task.\n\nRead owner labels, duplicate paths, source matches, and scan problems separately.\nThe `pnpm` owner names Skills linked from npm dependencies. Update these through pnpm, not skilld.\nDoctor never migrates or removes pnpm-owned links. `outdated --all` excludes them from replacement suggestions.\nExit code `1` with success data means some files could not be checked.\nUse `--check-sources` for up to 20 source candidates through skilld.dev.\nA name match alone never establishes provenance.\n\nThe interactive `skilld doctor` separates skills.sh installs with Tab.\nThe user can review migration with `m` or removal with `d`.\nEach action shows affected paths and requires confirmation.\nBackups include original files, lock metadata, and recovery instructions.\nUse the existing update workflow for skilld installs.\nHistorical commit searches and project-to-global deduplication are outside Doctor's current actions.\n\n## Account and configuration\n\nBefore authentication, account actions, or configuration, read [Account and configuration](references\u002Faccount-and-configuration.md).\n\n## Remove a Skill\n\nInspect the named Skill and its scope before removal:\n\n```sh\nskilld view \u003Cskill> --plain\nskilld remove \u003Cskill> --plain\n```\n\nAdd `--global` to both commands for a global Skill.\nRemove only the Skill and scope the user names.\nReport the removed Agent targets and whether recovery needs a reinstall.\n\n## Handle failures\n\nPreserve the original error code and message.\nDo not hide a failure with a fallback source or scope.\nDo not retry with `--direct` because it changes the source status.\n\nFor authentication errors, run `skilld auth status` before login.\nFor `AUTH_REQUIRED` from an account command, ask the user to run `skilld auth login`.\nFor `FORBIDDEN`, tell the user their account cannot make that change.\nFor target errors, inspect `agent.targets` and the requested `--agent` values.\nFor lockfile errors, preserve the lockfile and report its path.\nFor target conflicts, stop before overwriting existing files.\n\nIf a command partially completes, report every successful and failed Skill.\nNever claim success from generated commands that were not run.\n","# Use skilld CLI\n\nUse skilld CLI to find, load, install, inspect, restore, update, verify, and remove Skills.\n\nRun a Skill first. Install a Skill only when the user asks to keep it.\nIf the user asks to fork a Skill, copy its source before installing the local copy.\n\n## Use Agent output\n\nUse `--json` with `search`, `run`, and `update --check`.\nUse `--json` with `view` of a registry ref, and with every registry and account command below.\nTheir `data` is the skilld.dev answer. A command whose answer has no body returns `data: null`.\nUse `--json` with `sync` and `sync --check` for declared Skills.\nUse `doctor --json` for read-only Skill discovery and cleanup findings.\nThe remaining commands do not support JSON output.\nUse `--plain` when another command needs stable text.\n\nCheck the exit code before reading stdout.\nRead JSON success data only when `_tag` is `Success`.\nRead JSON failures from stderr.\nReport the error `code` and `message`.\nA failed `skilld run` ends with a `Next step:` line. In JSON, read `error.nextStep`.\nFollow that step. It says whether to retry, when, and what to tell the user.\nExit code `75` means a temporary failure, such as a network fault or a busy service.\nExit code `1` means the same command gives the same result.\nIf `skilld run` names files it loaded without, tell the user when the instructions need one.\n\nAn update check can exit with code 1 and return valid JSON.\nRead its update relations before treating that exit as a failure.\nNever parse formatted terminal output.\n\n## Sync declared Skills\n\nRead [Declared Skills](references\u002Fdeclared-skills.md) when a project needs explicit Skill requirements.\nUse `.skills\u002Fskilld.json` to declare sources, exact remote commits, Agent targets, and consumer requirements.\nRun `skilld sync --check --json` to find differences without installing or fetching remote bytes.\nExit code `1` with success data means sync is needed.\nRun `skilld sync` to install the full declaration in one transaction.\nUse `--global` for global Agent targets.\nUse `--adopt` only when migrating identical unmanaged symlinks.\nChanged unmanaged targets block sync.\nIf the project declares this Skill, use its declaration to refresh it.\nDo not install a second copy through another store.\nRequired Skills cannot be removed until their declaration releases the requirement.\n\nFor browser-controlled account login, use `skilld auth login --no-browser --plain`.\nOpen the printed URL in the intended signed-in browser while the command waits.\n\n## Search for a Skill\n\nRun a focused search:\n\n```sh\nskilld search \u003Cquery> --json\n```\n\nRead `data.items` before choosing a Skill.\nUse each item's `selector` for a Skill run.\nRefine the query when several Skills cover different tasks.\n\nDo not guess a selector from the Skill name.\nDo not install a search result before reading its description.\n\n## Run a Skill\n\nRun the selector returned by search:\n\n```sh\nskilld run \u003Cselector> --json\n```\n\nThe command prints SKILL.md and writes no Skill files.\nIt retains no remote Skill files after the command exits.\nRead the printed SKILL.md, then follow it for the current task.\nPrefer `skilld run` for a one-off task.\n\nRead `data.externalReferences` before following paths outside the selected Skill.\nThese notices identify references, not verified dependencies.\nFor a possible remote sibling Skill, `readArgv` pins its Repository and commit.\nRun that array only when the task and access permissions allow the read.\nNever add `--direct` to bypass an access failure.\nFor local references, resolve from the original Skill directory, not an Agent target copy.\nIf a reference is unresolved, identify its source before following it.\nInstall extra Skills only when the user asks to keep them.\n\nRead `data.files` for each supporting file's path, kind, and size.\nThe initial load prints no supporting file content.\nRead one only when the instructions name it:\n\n```sh\nskilld run \u003Cselector> --revision \u003Cdata.revision> --file \u003Cpath> --json\n```\n\nRun the exact `data.files[].readArgv` array when possible.\nIt contains the source, exact revision, file path, and `--json`.\nRepeat `--file` to read several files in one command.\n\nCheck `data.files[].readable` before you ask for a file.\nA file with `readable: false` never prints.\nIts `kind` is `executable` or `binary`.\nTell the user the Skill needs an install to use that file.\n\nRead `data.behaviors` before you follow the Skill.\nEach entry names one behavior, its `tier`, and the `path` and `line` where it appears.\nTell the user about every behavior before you act on it.\nAn empty list proves nothing. Patterns miss obfuscated code.\n\nA remote run stops with `BEHAVIOR_CONFIRMATION_REQUIRED` when the Skill has an `ask` behavior.\nskilld loaded nothing. Show the user every behavior in `error.message`.\nA match can carry a model reading, such as `(model reading: quoted example. REASON)`.\nShow it as a language model's reading. It never replaces the user's approval.\nIf the user approves, run the command at the end of the message and add `--json`.\nNever add `--allow` to any command without the user's approval in this session.\nIf the user declines, stop and load nothing.\n\nReport which Skill you ran and that skilld wrote no Skill files.\nRead `data.sourceStatus`, `data.origin`, and `data.revision`.\nA `verified` status covers where the Skill came from.\nIt does not cover what the instructions ask you to do.\nIf the status is `unverified`, tell the user before you follow the Skill.\n\n## Author a Skill from the user's own code\n\nThe skilld CLI has no generation command.\nAuthoring lives in skilld-maintained Skills that you run.\n\nRun the Skill that matches the request:\n\n```sh\nskilld run skilld-dev\u002Fskilld\u002Fgenerate-project-skill --json\nskilld run skilld-dev\u002Fskilld\u002Fgenerate-package-skill --json\nskilld run skilld-dev\u002Fskilld\u002Freview-skill --json\n```\n\nUse `generate-project-skill` for the current project or workspace.\nUse `generate-package-skill` for a package the user maintains.\nUse `review-skill` before the user publishes a Skill.\n\nRead the printed SKILL.md, then follow it for this project.\nWrite the draft where the user asks, and show the files for review.\nDo not replace an existing Skill until the user approves the draft.\nDo not install or publish a draft the user has not read.\n\nThe user owns the draft. They commit it to their own Repository.\n\n## List the Skills a Repository, curator, or collection names\n\nskilld.dev prints refs that name several Skills:\n\n- `OWNER\u002FREPOSITORY` names every Skill in one Repository.\n- `@LOGIN` names every Skill in one curator's collections.\n- `@LOGIN\u002FSLUG` names every Skill in one collection.\n\nRun one of these refs to list its Skills:\n\n```sh\nskilld run @LOGIN\u002FSLUG --json\n```\n\nThe command prints an index and loads no Skill.\nRead `data.items` for each Skill's `name`, `owner`, `repository`, `description`, and `selector`.\nRun the `data.items[].runArgv` array to load one Skill.\nPick the Skills the current task needs. Do not run every Skill in the index.\n\n## Read the registry\n\nRead one registry entry before you run or install it:\n\n```sh\nskilld view OWNER\u002FREPOSITORY\u002FSKILL --json\n```\n\nRead `data.owner`, `data.sourceUrl`, `data.sourceCommit`, and `data.description`.\nReport the author and the exact SKILL.md with the Skill.\n`skilld view` also takes `OWNER\u002FREPOSITORY`, `@LOGIN`, and `@LOGIN\u002FSLUG`.\nA bare name without `\u002F` or `@` shows an installed Skill.\n\nUse these commands when the user asks what exists, not for one task:\n\n```sh\nskilld browse \u003Cquery> --sort stars --json\nskilld trending --json\nskilld tracks --json\nskilld tracks \u003Cslug> --json\nskilld curators --json\n```\n\nUse `skilld search` to find a Skill for the current task.\nEach trending item has a `signal`. Its `kind` says why the Skill trends.\nReport that reason with the Skill.\nUse `--limit` and `--offset` to page. `data.total` counts every result.\n\nThe user can name a Repository the registry does not list. Ask skilld.dev to index it:\n\n```sh\nskilld index OWNER\u002FREPOSITORY --json\n```\n\nThe command waits about a minute.\nIf `data.status` is still `queued`, run the same command again later.\nAn `INDEX_FAILED` error names the reason.\n\n## Choose the source\n\nPrefer the exact `OWNER\u002FREPOSITORY\u002FSKILL` selector returned by Skill search.\nHosted selectors use immutable artifact delivery from an exact Git commit.\n\nUse a local path only for a Skill the user already controls:\n\n```sh\nskilld run .\u002Fskills\u002Fmy-skill --json\nskilld install .\u002Fskills\u002Fmy-skill\n```\n\nUse `--direct` only for an explicit public GitHub selector.\nDirect mode bypasses artifact delivery and gives the `unverified` source status.\nIt cannot access a private repository.\nNever add `--direct` merely to bypass a delivery failure.\n\nPrivate repository delivery requires a skilld.dev account and GitHub App access.\nIf private access fails, check authentication before changing the selector.\n\n## Install a Skill\n\nInstall a Skill when the user wants it in every session.\nInstall a Skill when it must run its own script.\nAn explicit install or fork request authorizes the requested project files.\nOtherwise, ask before installing.\n\nInstall the selector returned by search into the detected Agent target:\n\n```sh\nskilld install \u003Cselector>\n```\n\nThe default scope is the current project.\nProject installs update `.skills\u002Fskilld-lock.yaml` and selected Agent targets.\n\nInstall into global Agent targets:\n\n```sh\nskilld install \u003Cselector> --global\n```\n\nUse `--agent \u003Cagent>` when the user names an Agent target.\nRepeat `--agent` when the user names several Agent targets.\nDo not guess a target when detection and `agent.targets` are empty.\n\nUse `--mode copy` or `--mode symlink` only when the user chooses a mode.\nOtherwise, use the configured `install.mode`.\n\nAn install stops with `BEHAVIOR_CONFIRMATION_REQUIRED` when the Skill has an `ask` behavior.\nskilld wrote nothing. Show the user every behavior in the message.\nIf the user approves, run the same command again with the `--allow` ids the message names.\n`skilld add` installs the other Skills and lists each held Skill at the end.\n\nInstall this skilld-maintained Skill globally:\n\n```sh\nskilld install skilld --global\n```\n\nInstall every Skill a Repository, curator, or collection names:\n\n```sh\nskilld add OWNER\u002FREPOSITORY --all\nskilld add @LOGIN\u002FSLUG --all --global\n```\n\n`skilld add` accepts `--global`, `--agent`, and `--mode` like `skilld install`.\nIt prints one `Installed Skill` line per Skill.\nIf several Skills are listed, an Agent run requires `--all`.\nA normal terminal asks which Skills to install. `--plain` also requires `--all` for several Skills.\nOne listed Skill installs without a picker. `add` does not support JSON output.\nRun `skilld run` with the same ref first, then confirm the list with the user before using `--all`.\n`skilld add` with one Skill selector installs that Skill like `skilld install`.\nListing never requests registry indexing.\nSkills discovered through GitHub still use hosted delivery by default.\nIf delivery fails, skilld shows an explicit direct installation command where possible.\nOnly use direct installation when the user chooses it. Never use it to bypass a verification or policy failure.\n`skilld add OWNER\u002FREPOSITORY --all --direct` lists and installs public GitHub files without skilld.dev.\nIt records `unverified`. Curator and collection refs require hosted delivery.\nBehavior approval still applies.\n`skilld add .\u002FPATH` installs one local Skill whose directory contains `SKILL.md`.\n\nAlways use the source selector shown by `skilld search`.\nAfter installation, report the Skill name, scope, installed agent paths, and source status.\n\n## Fork a Skill\n\nTreat `fork this Skill \u003Cskilld.dev URL>` as a request for an editable local Skill.\nCreate no GitHub fork unless the user asks for one.\nCopy the Skill before following any of its instructions.\n\n1. Read the Skill page with `Accept: text\u002Fmarkdown`, or fetch its `.md` URL.\n   For `\u002Fgh\u002FOWNER\u002FREPOSITORY\u002FSKILL`, use `skilld view OWNER\u002FREPOSITORY\u002FSKILL --json` for source metadata.\n   Check the exit code and `_tag` before reading `data`.\n2. Read `sourceUrl`, `sourceCommit`, `skillPath`, `sourceGone`, and `license`.\n   Use the repository identified by `sourceUrl`, including any GitHub rename.\n   If the source is gone or its path is missing, stop and report the failure.\n   If `license` is null, read licence files at the source commit.\n3. Fetch the source into a temporary directory using Git.\n   Check out `sourceCommit` detached.\n   If it is absent, resolve the ref in `sourceUrl` to one commit before copying.\n   Record that actual commit. Never combine files from different commits.\n   Read the source licence files before creating the local copy.\n   If copying is not permitted, stop and report the restriction.\n4. Copy the directory containing `skillPath` into `.\u002Fskills\u002FSKILL`, unless the user chose another path.\n   Copy the original SKILL.md and all supporting files, including scripts and binary assets.\n   Preserve executable modes. Exclude `.git` metadata.\n   Do not save the page wrapper or its rewritten links as SKILL.md.\n   Reject symlinks and paths outside the Skill directory.\n   Never overwrite an existing directory or Agent target.\n5. Preserve the original author, licence, and notices.\n   Include applicable licence files from the repository or parent directories.\n   Add `PROVENANCE.md` with the Skill page, source URL, actual commit, original path, and licence.\n   If that file already exists, retain it and record provenance in a separate file.\n   Do not replace the original author's credit with the user's name.\n6. In the project root, run `skilld install .\u002Fskills\u002FSKILL --mode copy`.\n   First inspect the project lockfile and selected Agent target directories for this Skill name.\n   If it is already installed, stop before replacing it.\n   Use the detected Agent targets. Pass `--agent` only for targets the user selected.\n   This records a local source. Never install the upstream selector for a fork.\n   If target selection or installation fails, preserve the local copy and report the exact failure.\n7. Report the local path, source commit, and installed Agent targets.\n   After local edits, reinstall the same local path to refresh Agent targets.\n   Upstream updates must not replace the local copy.\n\nDo not publish or push the local copy unless the user asks.\n\n## Restore locked Skills\n\nRestore the current project from its lockfile:\n\n```sh\nskilld install\n```\n\nRestore global Skills from the global scope:\n\n```sh\nskilld install --global\n```\n\nA verified remote Skill restores its exact locked commit through artifact delivery.\nAn unverified remote Skill requires the recovery command shown by skilld.\nDo not convert a verified source to direct mode during recovery.\n\nNever delete a lockfile or Agent target to repair an install.\nPreserve the files and report the exact failure first.\n\n## Inspect installed Skills\n\n```sh\nskilld list\nskilld list --global\nskilld view \u003Cskill>\nskilld view \u003Cskill> --global\n```\n\nUse `list` to find installed names in one scope.\nUse `view` to inspect a Skill before any mutation.\nRead its path, source, source status, and Agent targets.\n\n## Check and apply updates\n\n```sh\nskilld update --check --json\nskilld update \u003Cskill>\nskilld update \u003Cskill> --global\n```\n\nUse `update --check --json` to inspect update relations without changing files.\nRead each `data.items[].relation._tag` before changing files.\nUse `update \u003Cskill>` only when the relation is `available`.\nTreat `current`, `pinned`, and `notTracked` as no action.\nIf the relation is `behind` or `diverged`, ask before changing files.\nIf the relation is `unavailable`, report `failure.code` and `failure.message`.\nTreat `unavailable` as unknown. Do not infer a newer commit.\n\nAn update stops with `BEHAVIOR_CONFIRMATION_REQUIRED` when the new version adds an `ask` behavior.\nskilld changed nothing. Show the user every behavior in the message.\nIf the user approves, run the same command again with the `--allow` ids the message names.\n\nUpdate one named Skill unless the user explicitly requests all updates.\nUse `--global` only for a Skill in the global scope.\nLeave `--interactive` to a human terminal session.\n\n## Verify source integrity\n\n```sh\nskilld verify \u003Cskill>\n```\n\nUse `verify` to check installed bytes against recorded source data.\nA successful check confirms provenance and integrity only.\nIt does not approve the Skill instructions.\n\nIf verification fails, do not hand edit a managed Skill.\nUse `view` to inspect its source before update or restore.\n\n## Report outdated and unmanaged Skills\n\nCheck the current scope:\n\n```sh\nskilld outdated --plain\n```\n\nCheck both scopes and every Agent target directory:\n\n```sh\nskilld outdated --all --plain\n```\n\nUse `outdated` for stale, unverified, local, and unmanaged Skill reports.\nRead every proposed command before using it.\nNever delete an unmanaged Skill unless the user names it for removal.\n\n## Review Skill files across projects\n\nRun `skilld doctor --json` to scan the user's home directory.\nPass explicit roots to narrow the scan or inspect files outside home.\nThe default scan excludes dependencies, caches, generated files, backups, and Git worktrees.\nUse `--exclude GLOB` to prune more paths.\nUse `--include-excluded` or `--include-worktrees` only when those files matter to the task.\n\nRead owner labels, duplicate paths, source matches, and scan problems separately.\nThe `pnpm` owner names Skills linked from npm dependencies. Update these through pnpm, not skilld.\nDoctor never migrates or removes pnpm-owned links. `outdated --all` excludes them from replacement suggestions.\nExit code `1` with success data means some files could not be checked.\nUse `--check-sources` for up to 20 source candidates through skilld.dev.\nA name match alone never establishes provenance.\n\nThe interactive `skilld doctor` separates skills.sh installs with Tab.\nThe user can review migration with `m` or removal with `d`.\nEach action shows affected paths and requires confirmation.\nBackups include original files, lock metadata, and recovery instructions.\nUse the existing update workflow for skilld installs.\nHistorical commit searches and project-to-global deduplication are outside Doctor's current actions.\n\n## Account and configuration\n\nBefore authentication, account actions, or configuration, read [Account and configuration](references\u002Faccount-and-configuration.md).\n\n## Remove a Skill\n\nInspect the named Skill and its scope before removal:\n\n```sh\nskilld view \u003Cskill> --plain\nskilld remove \u003Cskill> --plain\n```\n\nAdd `--global` to both commands for a global Skill.\nRemove only the Skill and scope the user names.\nReport the removed Agent targets and whether recovery needs a reinstall.\n\n## Handle failures\n\nPreserve the original error code and message.\nDo not hide a failure with a fallback source or scope.\nDo not retry with `--direct` because it changes the source status.\n\nFor authentication errors, run `skilld auth status` before login.\nFor `AUTH_REQUIRED` from an account command, ask the user to run `skilld auth login`.\nFor `FORBIDDEN`, tell the user their account cannot make that change.\nFor target errors, inspect `agent.targets` and the requested `--agent` values.\nFor lockfile errors, preserve the lockfile and report its path.\nFor target conflicts, stop before overwriting existing files.\n\nIf a command partially completes, report every successful and failed Skill.\nNever claim success from generated commands that were not run.\n",{"name":5,"license":14,"description":7,"compatibility":47},"Requires an agent with filesystem and shell access. Network access is needed for remote sources.","main","d19d25b79cfdf6bef33bd8f981f9dca46ffcffa2","skills\u002Fskilld",10,"https:\u002F\u002Fgithub.com\u002Fskilld-dev\u002Fskilld","https:\u002F\u002Fskilld.dev","https:\u002F\u002Fgithub.com\u002Fskilld-dev","Curated agent skills by humans.","2026-10-11 00:23:17",{"source":8,"skillName":5,"commands":58,"prompt":68,"zipAvailable":69},[59,62,65],{"tool":60,"command":61},"npx","npx skills add skilld-dev\u002Fskilld --skill skilld",{"tool":63,"command":64},"bunx","bunx skills add skilld-dev\u002Fskilld --skill skilld",{"tool":66,"command":67},"pnpm","pnpm dlx skills add skilld-dev\u002Fskilld --skill skilld","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fskilld-dev\u002Fskilld\u002Fblob\u002Fmain\u002Fskills\u002Fskilld\u002FSKILL.md",true,[71,76,82,86,90],{"path":72,"resourceType":72,"resourceTypeDesc":73,"sizeBytes":74,"sourceHtmlUrl":75},"LICENSE","许可证",1071,"https:\u002F\u002Fgithub.com\u002Fjiayaoqijia\u002Fcryptoskill\u002Fblob\u002Fmain\u002Fskills\u002Fmcp-servers\u002Fdanthamanvoiss-salesportl-professional-services-client-success-expansion\u002FLICENSE",{"path":77,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":80,"sourceHtmlUrl":81},"references\u002Faccount-and-configuration.md","REFERENCE","参考文档",2964,"https:\u002F\u002Fgithub.com\u002Fskilld-dev\u002Fskilld\u002Fblob\u002Fmain\u002Fskills\u002Fskilld\u002Freferences\u002Faccount-and-configuration.md",{"path":83,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":84,"sourceHtmlUrl":85},"references\u002Fdeclared-skills.md",4710,"https:\u002F\u002Fgithub.com\u002Fskilld-dev\u002Fskilld\u002Fblob\u002Fmain\u002Fskills\u002Fskilld\u002Freferences\u002Fdeclared-skills.md",{"path":87,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":88,"sourceHtmlUrl":89},"SOURCE.md",518,"https:\u002F\u002Fgithub.com\u002Fjiayaoqijia\u002Fcryptoskill\u002Fblob\u002Fmain\u002Fskills\u002Fmcp-servers\u002Fdanthamanvoiss-salesportl-professional-services-client-success-expansion\u002FSOURCE.md",{"path":91,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":92,"sourceHtmlUrl":93},"TRUST.auto.yaml",2217,"https:\u002F\u002Fgithub.com\u002Fjiayaoqijia\u002Fcryptoskill\u002Fblob\u002Fmain\u002Fskills\u002Fmcp-servers\u002Fdanthamanvoiss-salesportl-professional-services-client-success-expansion\u002FTRUST.auto.yaml",[95,107,127,149,163,175],{"skillNo":96,"name":97,"nameZh":6,"description":98,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":5,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":99,"categories":100,"keywords":102,"publishedAt":42},"01M6CHCPXXZHPGTWWN5WEX9G1E","local-skill","Fixture Skill for the Rust and WASIp2 command proof.","https:\u002F\u002Fgithub.com\u002Fskilld-dev\u002Fskilld\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fv3-rust\u002Flocal-skill\u002FSKILL.md",[101],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},[103,104,105,106],"local skill","rust","command","fixture skill",{"skillNo":108,"name":109,"nameZh":6,"description":110,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":5,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":111,"categories":112,"keywords":117,"publishedAt":42},"01M6CHCM988WKQT5X5R6WD0PG3","update-package-skill","Updates an existing package Skill for a new release of its npm or local package by testing only what changed since the version the Skill names. Use when a maintainer asks to refresh a package Skill after a release, or when a SKILL.md names an older tested version than the package.","https:\u002F\u002Fgithub.com\u002Fskilld-dev\u002Fskilld\u002Fblob\u002Fmain\u002Fskills\u002Fupdate-package-skill\u002FSKILL.md",[113,116],{"categoryNo":114,"categoryName":115,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000008","测试与质量",{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},[118,119,120,121,122,123,36,33,31,30,124,26,125,126],"update","package","update package skill","testing","release","npm","refresh","version","only",{"skillNo":128,"name":129,"nameZh":6,"description":130,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":5,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":131,"categories":132,"keywords":142,"publishedAt":42},"01M6CHCDKF6DETR87DJ60YE5A1","skilld-tui-design","Designs and verifies skilld CLI output, terminal pickers, and full-screen flows. Use for terminal colours, loading feedback, keyboard controls, layout, errors, or end-to-end CLI UX reviews.","https:\u002F\u002Fgithub.com\u002Fskilld-dev\u002Fskilld\u002Fblob\u002Fmain\u002Fskills\u002Fskilld-tui-design\u002FSKILL.md",[133,136,137],{"categoryNo":134,"categoryName":135,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000015","视觉设计",{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},{"categoryNo":138,"categoryName":139,"taxonomyType":140,"taxonomyTypeDesc":141},"01JBSEEDCAT000000000000032","设计师","PROFESSION","适用职业",[143,144,145,33,31,146,30,36,147,148,26,32],"skilld tui design","tui","design","feedback","ux","terminal",{"skillNo":150,"name":151,"nameZh":6,"description":152,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":5,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":153,"categories":154,"keywords":156,"publishedAt":42},"01M6CHCB4S8W1XVP30QK48PT7M","review-skill","Review an Agent Skill for valid structure, clear triggers, usable instructions, current evidence, and risky or unclear actions.","https:\u002F\u002Fgithub.com\u002Fskilld-dev\u002Fskilld\u002Fblob\u002Fmain\u002Fskills\u002Freview-skill\u002FSKILL.md",[155],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},[157,158,159,160,161,36,33,31,30,162],"review","review skill","instructions","valid","actions","evidence",{"skillNo":164,"name":165,"nameZh":6,"description":166,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":5,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":167,"categories":168,"keywords":170,"publishedAt":42},"01M6CHC8NP537XAZ94CPMZWXYA","generate-project-skill","Generate or update an Agent Skill from the observed workflows, boundaries, and conventions of one software project.","https:\u002F\u002Fgithub.com\u002Fskilld-dev\u002Fskilld\u002Fblob\u002Fmain\u002Fskills\u002Fgenerate-project-skill\u002FSKILL.md",[169],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},[171,172,173,174,118,36,33,31,30],"project skill","generate project skill","generate","workflows",{"skillNo":176,"name":177,"nameZh":6,"description":178,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":5,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":179,"categories":180,"keywords":182,"publishedAt":42},"01M6CHC5PMY75CEJXN6FWGRSVP","generate-package-skill","Generate or update an Agent Skill that teaches consumers one npm or local package the user maintains, including framework modules and wrappers. Tests each example against the installed version. Use when a maintainer asks for a package Skill, a SKILL.md for their library, or a Skill update after a release.","https:\u002F\u002Fgithub.com\u002Fskilld-dev\u002Fskilld\u002Fblob\u002Fmain\u002Fskills\u002Fgenerate-package-skill\u002FSKILL.md",[181],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},[183,173,119,184,118,122,185,123,186,36,33,31,30,187,26,125,188],"generate package skill","modules","framework","library","user","example"]