AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Skill Finder

skill-ramananani-agent-router-skill-finder · by RamanaNani

Browse and discover Claude Code skills, agents, and plugins from the marketplace — ranked by rating — that are NOT yet installed. Use when the user is new and has few/no specialists installed, asks "what plugins/skills are available for X", "find me a skill for Y", "what should I install", "browse the marketplace", or when agent-router finds no installed tool that fits. Returns a ranked shortlist…

No reviews yet
0 installs
8 views
0.0% view→install

Install

$ agentstack add skill-ramananani-agent-router-skill-finder

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-ramananani-agent-router-skill-finder)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
1mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Skill Finder? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Skill Finder — marketplace discovery

Where agent-router routes among tools you ALREADY have, skill-finder discovers tools you DON'T have yet. It searches the marketplace, ranks candidates by rating, and hands back the exact install command. Built for a new user whose ~/.claude/ is empty or thin.

When to use

  • A new user wants to know what's worth installing for their kind of work.
  • The user asks "find/browse skills for X", "what plugin does Y", "what should I install".
  • agent-router found nothing installed scoring ≥ 25 → fall through to here.

Procedure

1. Capture the need

One sentence: the domain + task (code review, testing, security, infra, data, research, design…). Note any language/framework constraint.

2. Don't recommend what they already have

Quickly scan installed tools (the session list, ~/.claude/, ./.claude/, and installed plugins) so the shortlist is all NEW installs. If a good specialist is already installed, say so and route to it instead (defer to agent-router).

3. Gather candidates — real ones only

Merge two sources:

  • Curated catalog: data/marketplace.json (verified entries + ratings).
  • Live search (preferred, for freshness):
  • GitHub: repositories matching claude code plugin|skill|agent,

sorted by stars (use the GitHub search tool or a web search).

  • Web: claude code skill OR plugin OR subagent.

For each candidate capture: name, what it does, a rating signal (GitHub stars or curated score), the source URL, and how to install it.

NEVER invent a repo, package, or star count. List only candidates that appear in the catalog or in a real search result, and include the URL so the user can verify.

4. Rank (0-100)

  • Relevance (0-50): match to the task/domain.
  • Rating (0-35): GitHub stars (log-scaled) or curated rating. Unknown → 10.
  • Maintenance (0-15): recent commits / not archived. Unknown → 7.

Drop anything below relevance 15 — discovery should stay on-topic.

5. Present the shortlist

Top 3-5 as a table: rank, name, what it does, rating (e.g. ★ 1.2k or score 80), source link. Then, for the top pick, show the exact install command:

/plugin marketplace add /
/plugin install @

or npx when the project ships an npm installer.

6. Offer to install

Installing plugins is done by the USER via /plugin … — you cannot run those slash commands for them. Lay out the 1-2 commands and offer to walk through it. If the pick is an npx installer, you may run it via Bash once the user confirms.

7. Log the discovery (internal dogfooding)

Append ONE JSONL line so gaps in the installed toolset are tracked over time:

mkdir -p ~/.claude/agent-router/logs
echo '{"ts":"'"$(date -u +%FT%TZ)"'","skill":"skill-finder","need":"","domain":"","recommended":"","source":"","rating":"","installed":false,"feedback":""}' >> ~/.claude/agent-router/logs/decisions.jsonl

Set installed to true once the user installs it. These rows reveal which capabilities the user repeatedly needs but doesn't have — the roadmap.

Rules

  • Honesty over coverage: a short list of real, linked tools beats a long invented one.
  • Always show the source URL and the install command — discovery is useless without them.
  • State the rating's basis (GitHub stars vs. curated score) so it isn't mistaken for gospel.
  • If live search is unavailable and the catalog is thin, say so and show what you have.
  • After install, hand off to agent-router to actually route the task.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.