mgiovani/ai-arsenal / skills/find-skills
find-skills
VerifiedSearch, install, update, and remove third-party Agent Skills from skills.sh or any GitHub/GitLab repo via the `npx skills` CLI. Use when the user wants to "find a skill", "install a skill from github", "search skills.sh", or add third-party capabilities to Claude Code/Cursor/other agents. Not for authoring new skills (see create-skill).
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 7.2 KB |
| evals/evals.json | config | 4.0 KB |
| evals/trigger-eval.json | config | 1.8 KB |
| references/commands.md | markdown | 6.6 KB |
| references/workflows.md | markdown | 6.4 KB |
evals/evals.json
View on GitHub{
"skill": "find-skills",
"evals": [
{
"id": "search-testing-skill",
"prompt": "Is there a skill for writing Playwright tests? I don't want to write one myself.",
"assertions": [
"Runs `npx skills find` with a query related to playwright or testing (e.g. `npx skills find playwright` or `npx skills find testing`)",
"Does not run `npx skills add` or install anything -- the user only asked to search",
"Reports the search results back to the user instead of silently installing a result"
]
},
{
"id": "global-vs-project-scope",
"prompt": "Install a general commit-message-writing skill from someone's repo so I have it in every project I work on.",
"assertions": [
"The `npx skills add` command includes the `-g`/`--global` flag, because the user explicitly said 'every project I work on'",
"Targets a specific agent with `-a claude-code` (or asks which agent) rather than blindly accepting whatever the CLI defaults to",
"Does not install this skill at project scope (no plain `.claude/skills/` install without `-g`)"
]
},
{
"id": "ambiguous-scope-no-guess",
"prompt": "Add the react-patterns skill from acme/react-skills.",
"assertions": [
"Since nothing in the request signals project-specific vs personal/global use, the agent asks the user whether to install at project or global scope before running `npx skills add`",
"No `npx skills add` command (or any other mutating command) is run before the user answers the scope question",
"The final message poses the scope question clearly and stops there -- it does not proceed to install a guessed scope in the same turn",
"No state-mutating `npx skills` command of any kind (add, remove, update, or an install run against a fabricated local/mock repo as a substitute when acme/react-skills can't be resolved) appears anywhere in the transcript before the scope question is asked"
]
},
{
"id": "single-skill-removal-confirmation",
"prompt": "Remove the docs-diagram skill, I don't use it anymore.\n\nSANDBOX SETUP (do this first, exactly): create a project dir with `.claude/skills/docs-diagram/SKILL.md`, `.claude/skills/review-code/SKILL.md`, and `.claude/skills/i18n-check/SKILL.md` (each a minimal valid skill with name + description frontmatter), so that listing installed skills genuinely finds docs-diagram among others. Work from inside that project dir. Then handle the user's request above as a live turn.",
"assertions": [
"Before removing anything, the agent runs `npx skills list` (or equivalent, checking both project and global scope) to confirm docs-diagram is actually installed and where, and surfaces that to the user",
"The final message asks the user to explicitly confirm removing docs-diagram and stops there -- it does not run `npx skills remove docs-diagram` or `npx skills rm docs-diagram` in this same turn, even though the user named the skill precisely",
"No state-mutating `npx skills` command (remove, rm, add, update) appears anywhere in the transcript before the confirmation request"
]
},
{
"id": "bulk-removal-confirmation",
"prompt": "Clean up all my installed skills, I want a fresh start.",
"assertions": [
"Before proposing or running a destructive bulk-removal command (`npx skills remove --all` or equivalent `--skill '*' --agent '*'`), the agent surfaces what is currently installed (e.g. via `npx skills list`) and asks for explicit confirmation",
"No skill-removal command is executed in this turn without the user's affirmative confirmation being present in the conversation",
"The final message asks the user to confirm the bulk removal and goes no further -- it does not perform the removal preemptively",
"No state-mutating `npx skills` command of any kind (remove, rm, --all, add, update) appears anywhere in the transcript before the confirmation request is made"
]
}
]
}
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |