ClawTrust LogoClawTrust
Skill Finder (Find ClawHub skills + Search Skills.sh)

Skill Finder (Find ClawHub skills + Search Skills.sh)

by ivangdavila · v1.0.0

Productivity
ClawHub
8.5
/ 10
1 evaluations
3.9k Downloads

Overview

Helps the agent systematically discover, compare, and recommend installable skills from both ClawHub and Skills.sh based on the user’s needs, preferences, and past feedback, while enforcing safe, opt‑in installation practices.

Key Advantages

1.Dual‑ecosystem search across ClawHub and Skills.sh with a configurable source mode (both / clawhub / skills.sh).
2.Need-driven querying (search by capability gap and task intent rather than just exact skill names).
3.Structured evaluation of candidates using criteria like description clarity, downloads, last update, and author/repo reputation.
4.Persistent local memory (source mode, explicit preferences, liked/passed skills, search history) for increasingly personalized recommendations.
5.Clear recommendation format that ranks top options, explains tradeoffs, and highlights a single best fit when possible.

 Strong safety posture: installs are always opt‑in, no force flags or auto-“-

Use Cases

  • When a user asks how to accomplish a task and may be better served by installing a dedicated skill (e.g., PDF editing, data extraction, code review).
  • When the agent detects a capability gap or a repetitive workflow and wants to see if an existing skill can cover or automate it.
  • When a user requests a safer, more actively maintained, or more popular alternative to an existing skill.
  • When a workspace owner wants to curate or upgrade the local skill stack by comparing options across ClawHub and Skills.sh.
  • When migrating or upgrading skills and needing to re-evaluate current choices against newer or better-maintained options.

Evaluation Scores

8.5
/ 10
Reliability
7.8
Functionality
9.0
Usability
8.5
Safety
9.2
Performance
7.5
Compatibility
8.0

Based on 1 evaluation · Latest: 3/19/2026

Download Trend

Loading...

Evaluation History (1)

8.5/103/19/2026
▼
OS: darwin-x64LLM: google/gemini-2.5-flash
**Quick judgment:** This is a strong, well-structured meta-skill for discovering and selecting other skills. It is particularly valuable in installations where users frequently hit capability gaps or want to optimize their toolchain, and where both ClawHub and Skills.sh are available. **What it does well** - Searches **both ClawHub and Skills.sh** (by default) and compares results instead of treating them in isolation. - Focuses on **user needs and workflows** rather than raw name matching (e.g., “PDF edit” vs just “PDF”). - Applies **explicit evaluation criteria** (downloads, update recency, description clarity, reputation, install friction) before recommending. - Maintains **local memory** in `~/skill-finder/` for source mode, preferences, liked/passed skills, and optional search history, enabling more tailored suggestions over time. - Enforces **safe installation practices**: no auto-installs, no `-y` auto-confirmation, no force-install flags, and attention to scanner warnings. - Provides a **clear recommendation format** with a “best fit,” tradeoffs, and explicit install commands per source. **Key risks & limitations** - **External dependencies:** Requires `npx clawhub` and `npx skills` along with network access; breakage or schema changes in either ecosystem can reduce reliability. - **Environment assumptions:** Assumes a filesystem with `~/skill-finder/` available for memory; non-standard or restricted environments may need adaptation. - **Latency & performance:** Searches and inspections rely on external CLIs/registries, so responsiveness depends on network and registry performance. - **Complexity for novices:** Dual-registry explanation, source-mode choices, and evaluation criteria may feel heavy for very simple, one-off tasks. **Recommended scenarios** - Workspaces where users **frequently ask for new capabilities** (“Is there a skill for X?”, “Can we do this faster?”). - Teams actively **curating or evolving their skill stack**, needing structured comparisons (downloads, maintenance, safety). - Security- or reliability-conscious setups where **blind installs are not acceptable**, and each skill must be justified. - Environments that plan to **leverage both ClawHub and Skills.sh** rather than committing to a single registry.

Comments (0)

Post a Comment

No comments yet. Be the first!