docs-seeker_skill

Overview

This skill offers script-first documentation discovery using the llms.txt standard and context7.com URL patterns. It automates detection, fetching, and analysis of library and framework docs so you can find API references, GitHub repo content, and technical guides without manually building URLs. The workflow is optimized for fast topic queries and comprehensive library searches.

How this skill works

Run the provided scripts in sequence: detect-topic to classify the query and extract library/topic, fetch-docs to construct context7.com URLs and retrieve llms.txt, and analyze-llms-txt to categorize results and recommend agent distribution. Scripts handle fallback chains (topic → general → repo) and error conditions automatically and return JSON or llms.txt output. Environment configuration is loaded from .env locations so scripts run with zero-token overhead.

When to use it

• Looking up a specific API or feature within a library (topic-specific search)
• Gathering comprehensive documentation for a whole library or framework (general search)
• Analyzing a GitHub repository or codebase for available docs and references
• Preparing distributed agent workflows with recommended URL/agent distribution
• Automating doc discovery in CI pipelines or developer tooling

Best practices

• Always run scripts in the prescribed order: detect-topic → fetch-docs → analyze-llms-txt
• Use topic-specific queries for fastest results and fewer URLs
• For broad library searches, expect more URLs and follow analysis recommendations for agent distribution
• Keep your .env configuration consistent across environments to ensure predictable fallback behavior
• Read analysis JSON to decide between single-agent, phased, or multi-agent strategies

Example use cases

• Quickly find how to use a UI component (e.g., date picker) in a specific library using a topic query
• Collect comprehensive Next.js documentation links for onboarding or research
• Run repository analysis as a fallback when topic or general searches miss content
• Automate documentation fetching in a CI job to produce llms.txt and then feed it to downstream agents
• Use analyze-llms-txt output to decide how many parallel agents to deploy for processing documentation