Skill Seekers Plugin¶

Generate Nexus skills from documentation URLs using AI.

Overview¶

The Skill Seekers plugin automates skill creation from online documentation:

Scrape documentation from any URL
Generate SKILL.md files automatically
Import existing SKILL.md files
Batch process multiple URLs
Auto-extract keywords and metadata

Installation¶

pip install nexus-plugin-skill-seekers

Commands¶

generate¶

Generate a skill from a documentation URL:

nexus skill-seekers generate <url> [options]

Options: - --name TEXT: Skill name (auto-generated if not provided) - --tier TEXT: Target tier (agent, tenant, system). Default: agent - --output-dir TEXT: Output directory for SKILL.md file

Examples:

# Generate from URL
nexus skill-seekers generate https://docs.stripe.com/api

# Specify custom name
nexus skill-seekers generate https://docs.example.com/api --name example-api

# Save to file only (don't import to Nexus)
nexus skill-seekers generate https://docs.example.com --output-dir ./skills

# Generate to tenant tier
nexus skill-seekers generate https://docs.example.com --tier tenant

What it does: 1. Fetches documentation from URL 2. Extracts main content 3. Generates skill name (if not provided) 4. Creates SKILL.md with frontmatter 5. Imports to Nexus (if available) 6. Optionally saves to file

Generated SKILL.md structure:

---
name: stripe-api
version: 1.0.0
description: Skill generated from https://docs.stripe.com/api
author: Skill Seekers
created: 2025-01-26T12:00:00Z
source_url: https://docs.stripe.com/api
tier: agent
---

# Stripe Api

## Overview

This skill was automatically generated from documentation at https://docs.stripe.com/api.

## Description

[Extracted content...]

## Keywords

api, data, request, response, object...

import¶

Import an existing SKILL.md file into Nexus:

nexus skill-seekers import <file-path> [options]

Options: - --tier TEXT: Target tier (agent, tenant, system). Default: agent - --name TEXT: Override skill name (uses filename if not provided)

Examples:

# Import skill file
nexus skill-seekers import ./my-skill.md

# Import to specific tier
nexus skill-seekers import ./team-skill.md --tier tenant

# Import with custom name
nexus skill-seekers import ./skill.md --name custom-name

batch¶

Generate multiple skills from a URLs file:

nexus skill-seekers batch <urls-file> [options]

Options: - --tier TEXT: Target tier for all skills. Default: agent

URLs file format:

# Lines starting with # are comments
https://docs.stripe.com/api stripe-api
https://docs.github.com/api github-api
https://docs.example.com/guide example-guide

Format: <url> [optional-name]

Example:

# Create urls.txt
cat > urls.txt << 'EOL'
https://docs.stripe.com/api stripe-api
https://docs.github.com/rest github-rest-api
https://docs.anthropic.com/claude claude-api
EOL

# Batch generate
nexus skill-seekers batch urls.txt

# Generate to tenant tier
nexus skill-seekers batch urls.txt --tier tenant

Output:

Processing 3 URLs...
Generating skills... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% 0:00:10
✓ Generated 3 skills

list¶

List all generated skills:

nexus skill-seekers list

Output:

┏━━━━━━━━┳━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Tier   ┃ Name         ┃ Path                                ┃
┡━━━━━━━━╇━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ Agent  │ stripe-api   │ /workspace/.nexus/skills/stripe-api │
│ Agent  │ github-api   │ /workspace/.nexus/skills/github-api │
│ Tenant │ team-skill   │ /shared/skills/team-skill           │
└────────┴──────────────┴─────────────────────────────────────┘

Use Cases¶

1. API Documentation Skills¶

Generate skills from API documentation:

# Stripe API
nexus skill-seekers generate https://docs.stripe.com/api --name stripe-api

# GitHub API
nexus skill-seekers generate https://docs.github.com/rest --name github-api

# Use the skill
nexus skills info stripe-api

2. Framework Documentation¶

Create skills from framework docs:

# Next.js
nexus skill-seekers generate https://nextjs.org/docs

# FastAPI
nexus skill-seekers generate https://fastapi.tiangolo.com

3. Library References¶

Build skill library from multiple sources:

# Create batch file
cat > libraries.txt << 'EOL'
https://pandas.pydata.org/docs pandas
https://numpy.org/doc numpy
https://docs.python.org/3/library python-stdlib
EOL

# Batch generate
nexus skill-seekers batch libraries.txt

4. Knowledge Base Migration¶

Import existing documentation:

# Generate from URLs
nexus skill-seekers generate https://internal-docs.company.com/api

# Or import existing markdown
nexus skill-seekers import ./existing-docs/*.md

Python API¶

Use the plugin programmatically:

from nexus import connect
from nexus.plugins import PluginRegistry

nx = connect()
registry = PluginRegistry(nx)
registry.discover()

plugin = registry.get_plugin("skill-seekers")

# Generate from URL
await plugin.generate_skill(
    url="https://docs.example.com/api",
    name="example-api",
    tier="agent"
)

# Import skill file
await plugin.import_skill(
    file_path="./my-skill.md",
    tier="tenant"
)

# List skills
await plugin.list_skills()

Workflow with Other Plugins¶

With Firecrawl Plugin¶

Use Firecrawl for better scraping:

# Scrape with Firecrawl (handles JS, anti-bot)
nexus firecrawl scrape https://docs.example.com/api --output api.md

# Import as skill
nexus skill-seekers import api.md --name example-api

With Anthropic Plugin¶

Generate and upload to Claude:

# Generate skill
nexus skill-seekers generate https://docs.stripe.com/api --name stripe-api

# Upload to Claude
nexus anthropic upload-skill stripe-api

Implementation Details¶

Content Extraction¶

The plugin uses BeautifulSoup for HTML parsing: 1. Fetches URL content 2. Removes scripts and styles 3. Extracts text content 4. Cleans whitespace 5. Generates structured SKILL.md

Name Generation¶

Auto-generates names from URLs:

https://docs.stripe.com/api → stripe-api
https://github.com/docs/rest → rest
https://example.com/guide/auth → auth

Rules: - Uses last path component or domain - Replaces special chars with - - Removes consecutive dashes

Keyword Extraction¶

Simple keyword extraction algorithm: 1. Finds common technical terms 2. Filters for programming/API keywords 3. Returns top 10 unique keywords

Limitations¶

Current Implementation¶

Basic content extraction (no LLM)
Simple keyword extraction
Single-page scraping only
No JavaScript rendering

Future Enhancements¶

Planned improvements: - LLM-powered content summarization - Multi-page documentation handling - Better keyword extraction with NLP - JavaScript rendering support - Schema extraction for APIs

Comparison with Firecrawl¶

Feature	Skill Seekers	Firecrawl
JavaScript rendering	❌ No	✅ Yes
Anti-bot detection	❌ No	✅ Yes
Multi-page crawl	❌ No	✅ Yes
Skill generation	✅ Yes	❌ No
Free	✅ Yes	⚠️ Limited
Dependencies	Minimal	Requires API key

Recommendation: Use Firecrawl for scraping, Skill Seekers for skill generation.

Troubleshooting¶

Failed to Fetch Documentation¶

Check URL accessibility:

curl -I https://docs.example.com

Some sites may block scraping.

Empty Content¶

The site might be JavaScript-heavy. Use Firecrawl instead:

nexus firecrawl scrape https://example.com --output content.md
nexus skill-seekers import content.md

Skill Already Exists¶

Remove or rename existing skill:

# Remove from Nexus
nexus skills delete old-skill

# Or use different name
nexus skill-seekers generate https://url --name new-name

Source Code¶

GitHub: nexus-plugin-skill-seekers
PyPI: nexus-plugin-skill-seekers