richlander/dotnet-inspect/package-skills
Claude Code and agent skill by richlander/dotnet-inspect
- community
Checking that the SKILL.md file is still in the repository.
Discover and use version-matched agent skills from NuGet packages, and persist selected skills in a repository only when the user explicitly requests it.
Install
npx vskill@latest install richlander/dotnet-inspect/package-skillsSKILL.md
dotnet-inspect: Discover and Load NuGet Package Skills
Discover version-matched package skills, inspect their inventory, and load the ones relevant to the current task. This default workflow is agent-driven and does not change the repository. Persist skills only when the user explicitly asks for repository installation. Use the recipes below directly; for unfamiliar query or payload rules, use -D, command help or explain "<question>" instead of guessing arguments.
Use this workflow to reach the right package-authored skill efficiently. Prefer the restored project's inventory or an exact package pin. For an unknown package ID, use find --packages, not API-symbol Find.
Default: use skills without changing the repository
Restore or build first when dependencies changed; project only reads the existing project.assets.json.
First ask whether the restored direct dependencies expose any skills:
If the count is nonzero, ask for the inventory. It includes the resolved package version, package-relative path, skill name, and description:
Use this view first because its package versions match the code. Compare the descriptions with the task and code, then request each relevant skill by its displayed row:
For a known positional choice, select the complete inventory row first. --row then addresses the selected sequence:
Request several skills as a group by issuing one independent command for each selected row. Keep each result separate; --raw intentionally carries no multi-document boundary.
The agent may perform this entire discovery and loading workflow without asking the user. Reading package guidance changes agent context, not repository state. Treat every inventory row as a candidate; do not use the first row as a proxy for the package.
For a package that is not restored in the project, use an exact package coordinate. First ask whether it has skill documents:
Then list the paths and inspect the YAML header of each candidate row before requesting the full document. Together, those headers form the package-only inventory:
Do not use an unpinned find --packages when the repository consumes a specific version.
dotnet-inspect validates restored-project skill names and descriptions. In that normalized inventory, YAML values that require containment are reported as [Text omitted: required containment]. A selected YAML header or full skill document that requires containment is replaced in full by the same text through stdout, structured output, and file destinations. Reversible visible escape spellings may still disambiguate literal package content; they are not instructions to decode before use. A NuGet dependency is provenance, not proof that agent instructions are safe.
User-requested workflow: persist repository skills
Follow this workflow only when the user explicitly requests repository installation. It changes tracked state and requires a merged pull request to persist.
The target repository owns the installation regime. Inspect its contributor instructions, existing skill layout, and active agent harness before choosing a destination. Common roots include skills/, .agents/skills/, .github/skills/, and .claude/skills/.
dotnet-inspect project ... -S Skills applies the same checks and refuses missing or noncompliant metadata. Do not recover a refused skill by sanitizing or renaming it. For an accepted skill, preserve that validated name as the leaf directory under the repository-selected root.
For example, Markout itself keeps its skills under skills/, so its output formats skill belongs at skills/markout-output-formats/SKILL.md. Create that directory and confirm the package-local row from the exact package coordinate. Then either redirect the contained stdout payload:
Or ask dotnet-inspect to write the same contained payload:
For authored package content, --prefer-rendered-urls retains the existing links instead of normalizing GitHub file links to fetchable form. Preserve each packaged document as its own skill rather than combining, renaming, or decoding contained text. If the skill references sibling scripts, references, or assets, inspect the package subtree and persist those beside SKILL.md too.
Do not duplicate the same skill into several roots; harnesses may discover duplicate names and the copies will drift.
Keep skills version-matched
Commit selected skills with the dependency change. Record the package id, resolved version, and package-relative path in the commit or pull request. On every package upgrade, rerun the project inventory, compare every installed skill with its new packaged counterpart, add newly relevant focused skills, and remove a skill only when the code no longer uses the behavior it covers.
Omit -o/--output to review the document on stdout before writing it. Load dotnet-inspect skill private-feeds when exact reacquisition needs custom sources or credentials. Use a local .nupkg path in place of Package@Version for an unpublished package canary.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/package-skills/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/richlander/dotnet-inspect/package-skills)