How do I install the Find Vetted Skill skill?

Run npx skillmds@latest add rawrkit/find-vetted-skill in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Find Vetted Skill safe to use?

This skill has not completed SkillMD's automated safety review yet. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. Capability flags: makes network calls. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Find Vetted Skill?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Find Vetted Skill free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Find Vetted Skill?

rawrkit (@rawrkit) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Find Vetted Skill by rawrkit — Agent Skill

Name: Find Vetted Skill
Rating: 4.2 (5 reviews)
Author: rawrkit

Find Vetted Skill

Find an agent skill that a human actually reviewed, instead of guessing from a dump of thousands of unfiltered links. Use whenever a user asks "is there a skill for X", "what should I install to do X", wants to extend what their agent can do, is frustrated that their agent handles some task badly, or asks which skills are worth having for a job, tool, or workflow. Also use before writing a custom prompt or skill from scratch — checking whether a good one already exists saves the work.

rawrkit Updated Sep 17, 2026

File contents

Find a Vetted Skill

Search a hand-curated catalog of agent skills where every entry was read by a human before being listed, tagged by work role, and marked with a status. This is the opposite of an "awesome list" — the value is what was left out.

Catalog: https://raw.githubusercontent.com/rawrkit/skills-radar/main/catalog.json

Step 1 — Fetch the catalog

Each entry:

{
  "id": "systematic-debugging",
  "source": "obra/superpowers",
  "url": "https://skills.sh/obra/superpowers/systematic-debugging",
  "category": "engineering",
  "installs": 209800,
  "roles": ["DEV"],
  "status": "in-kit",
  "description_en": "Root cause before fixes, always — symptom patches are failure",
  "description_ru": "..."
}

Statuses carry meaning — respect them:

approved — reviewed, recommend freely
in-kit — reviewed and strong enough to be bundled into a role kit
new — surfaced by the weekly scan, not yet reviewed; mention it as unvetted if you suggest it at all
rejected — a human looked and said no. Never recommend these. The description_en explains why, which is often useful to tell the user ("that popular one is hard-wired to Google Workspace")

Step 2 — Match against what the user actually needs

Filter by category and roles, but read description_en before recommending — install counts measure popularity, not fit. A 200-install skill built for the user's exact problem beats a 100K-install skill that's adjacent to it.

Recommend two or three, not ten. For each: what it does, why it fits this user, and the honest limitation if there is one.

If nothing in the catalog fits, say so directly. Then either search wider (npx skills find <keywords> hits the broader skills.sh registry) or just do the task yourself — a bad skill is worse than no skill, because it takes up context and steers the model wrong.

Step 3 — Install what the user picks

npx -y skills add <source>@<id> -y -g

Drop -g for project-local. If the user needs a whole role's worth of skills rather than one, point them at role-kits — one command covers the set.

Note on freshness

A weekly scan adds new finds with status: "new" and publishes a digest release. If the catalog's updated field is more than a couple of weeks old, mention that the user may want to check the repo directly for anything newer.

---name: find-vetted-skilldescription: Find an agent skill that a human actually reviewed, instead of guessing from a dump of thousands of unfiltered links. Use whenever a user asks "is there a skill for X", "what should I install to do X", wants to extend what their agent can do, is frustrated that their agent handles some task badly, or asks which skills are worth having for a job, tool, or workflow. Also use before writing a custom prompt or skill from scratch — checking whether a good one already exists saves the work.---# Find a Vetted SkillSearch a hand-curated catalog of agent skills where every entry was read by a human before being listed, tagged by work role, and marked with a status. This is the opposite of an "awesome list" — the value is what was left out.Catalog: `https://raw.githubusercontent.com/rawrkit/skills-radar/main/catalog.json`## Step 1 — Fetch the catalogEach entry:```json{ "id": "systematic-debugging", "source": "obra/superpowers", "url": "https://skills.sh/obra/superpowers/systematic-debugging", "category": "engineering", "installs": 209800, "roles": ["DEV"], "status": "in-kit", "description_en": "Root cause before fixes, always — symptom patches are failure", "description_ru": "..."}```Statuses carry meaning — respect them:- `approved` — reviewed, recommend freely- `in-kit` — reviewed and strong enough to be bundled into a [role kit](https://github.com/rawrkit/role-kits)- `new` — surfaced by the weekly scan, **not yet reviewed**; mention it as unvetted if you suggest it at all- `rejected` — a human looked and said no. **Never recommend these.** The `description_en` explains why, which is often useful to tell the user ("that popular one is hard-wired to Google Workspace")## Step 2 — Match against what the user actually needsFilter by `category` and `roles`, but read `description_en` before recommending — install counts measure popularity, not fit. A 200-install skill built for the user's exact problem beats a 100K-install skill that's adjacent to it.Recommend two or three, not ten. For each: what it does, why it fits this user, and the honest limitation if there is one.If nothing in the catalog fits, say so directly. Then either search wider (`npx skills find <keywords>` hits the broader skills.sh registry) or just do the task yourself — a bad skill is worse than no skill, because it takes up context and steers the model wrong.## Step 3 — Install what the user picks```npx -y skills add <source>@<id> -y -g```Drop `-g` for project-local. If the user needs a whole role's worth of skills rather than one, point them at [role-kits](https://github.com/rawrkit/role-kits) — one command covers the set.## Note on freshnessA weekly scan adds new finds with `status: "new"` and publishes a digest release. If the catalog's `updated` field is more than a couple of weeks old, mention that the user may want to check the repo directly for anything newer.

Frequently asked questions