home / skills / ontos-ai / skills-evaluator / skills-sh-searcher

skills-sh-searcher skill

safe

This skill helps you locate and install skills from the skills.sh marketplace, streamlining PDF, web scraping, and data processing tasks.

npx playbooks add skill ontos-ai/skills-evaluator --skill skills-sh-searcher

Review the files below or copy the command above to add this skill to your agents.

Files (4)

SKILL.md

3.6 KB

---
name: skills-sh-searcher
description: Search and download skills from skills.sh ecosystem. Use when user needs to find and install skills from the skills.sh marketplace for specific tasks like PDF processing, document editing, web scraping, etc.
---

# Skills.sh Searcher

This skill searches and downloads skills from the [skills.sh](https://skills.sh) marketplace, which is the largest directory of AI agent skills.

## Core Functionality

1. **Search Skills**: Search skills by keywords from skills.sh leaderboard
2. **Download Skills**: Download skill packages to `.opencode/skill/` directory
3. **Convert to Standard Format**: Convert skills.sh data to standard skill_card format
4. **Auto-Install**: Make skills immediately available for opencode

## Usage

**Trigger**: `/skills-sh-searcher <keywords>` or "Search skills.sh for PDF skills"

### Search Examples

```bash
# Search for PDF-related skills
keywords: ['pdf', 'document', 'file']

# Search for web scraping skills
keywords: ['scraping', 'web', 'html', 'bs4']

# Search for data processing skills
keywords: ['data', 'csv', 'json', 'excel']
```

### Download Examples

```bash
# Download specific skill
skills-sh-searcher --download anthropics/skills/pdf

# Download multiple skills
skills-sh-searcher --download anthropics/skills/docx
skills-sh-searcher --download anthropics/skills/xlsx
```

## Skills Available on skills.sh

Based on the skills.sh leaderboard, available skills include:

### Document Processing Skills
- `anthropics/skills/pdf` (923 installs) - PDF processing
- `anthropics/skills/docx` (735 installs) - Word document processing
- `anthropics/skills/xlsx` (773 installs) - Excel spreadsheet processing
- `anthropics/skills/pptx` (772 installs) - PowerPoint presentation processing

### Web Development Skills
- `anthropics/skills/frontend-design` (3.9K installs) - Frontend design
- `anthropics/skills/webapp-testing` (662 installs) - Web app testing
- `anthropics/skills/mcp-builder` (618 installs) - MCP server builder

### Data Processing Skills
- `anthropics/skills/webapp-testing` - Web scraping
- Various data transformation skills

### Testing Skills
- `anthropics/skills/webapp-testing` - Testing frameworks
- `anthropics/skills/test-driven-development` - TDD patterns

### Development Tools
- `anthropics/skills/skill-creator` (2.5K installs) - Skill creation guide
- Various IDE and coding patterns

## Implementation Workflow

1. **Fetch Skills from skills.sh**
   - Access skills.sh homepage
   - Parse skill information (name, installs, URL, tags)
   - Filter by keywords if provided

2. **Download Skills to .opencode/skill/**
   - Get skill details from GitHub
   - Create standardized SKILL.md
   - Copy to `.opencode/skill/` directory

3. **Quality Filtering**
   - Filter by install count (minimum threshold)
   - Filter by relevance to keywords
   - Return top matches

## Resources

- `scripts/fetch_skills_sh.py` - Fetch and parse skills.sh leaderboard
- `scripts/download_skill.py` - Download skill from GitHub to .opencode/skill/
- `references/skills_catalog.json` - Cache of available skills

## Best Practices

- Always check install count to find popular, reliable skills
- Use specific keywords for better matches
- Download skills with high install counts first
- Verify downloaded skills before using

## Notes

- skills.sh uses the format `owner/repo/skill_name` for skill identification
- Skills are installed using `npx skills add <owner/repo>` command
- This skill automates that process by:
  1. Finding the right skill
  2. Downloading it from GitHub
  3. Converting to standard SKILL.md format
  4. Placing in `.opencode/skill/` directory

Overview

This skill searches and downloads skills from the skills.sh marketplace and makes them available locally for immediate use. It finds relevant agent skills by keyword, ranks results by popularity and relevance, and fetches selected packages into a local skills directory. The skill converts marketplace metadata to a standard skill card format so downloaded skills integrate with your agent environment.

How this skill works

The skill queries the skills.sh leaderboard and parses each entry for name, installs, URL, and tags. It filters and ranks matches by keyword relevance and minimum install count, then downloads chosen skill packages from their source (e.g., GitHub). Downloaded packages are converted into a standard skill card format and placed in the local .opencode/skill/ directory so they can be installed or activated immediately.

When to use it

When you need a ready-made skill to handle tasks like PDF processing, document editing, or web scraping.
When you want to discover popular or niche skills by keyword instead of searching manually on the marketplace.
When you want to bootstrap an agent by adding community-tested skills quickly.
When you need to evaluate or compare multiple skills before integrating one into a workflow.

Best practices

Use specific keywords (filetypes, task names, or libraries) to narrow search results.
Prefer higher install counts for more mature and well-tested skills.
Review downloaded skill metadata and README-equivalent docs before running in production.
Download and test skills in an isolated environment before enabling them for live agents.
Keep a local cache of downloaded skill metadata to speed repeated searches.

Example use cases

Search for and install PDF extraction and annotation skills for document pipelines.
Find web scraping skills that use common libraries (e.g., BeautifulSoup) to collect site data.
Pull Excel and CSV transformation skills to automate data cleaning and reporting.
Rapidly assemble a set of frontend-design and testing skills for automated UI validation.
Evaluate multiple document-processing skills to choose the best fit for an ingestion pipeline.

FAQ

How does the skill choose which results to show first?

Results are ranked by keyword relevance and install count; you can set a minimum install threshold to filter out low-popularity items.

Where are downloaded skills stored?

Downloaded packages are placed in the local .opencode/skill/ directory and formatted into a standard skill card for immediate use.