Trending on GitHub

This tool transforms technical books and documents into unified agent skills for GitHub Copilot CLI, Amp, or Claude Code. It processes input files to distill content, achieving 24x to 51x token reduction compared to dumping full text into context. The result is a ready-to-study and reference skill that integrates directly into your AI coding workflow.

• Converts PDFs or document folders into agent skills for Claude Code and Copilot CLI.
• Reduces token usage by 24-51x compared to full-context injection for single questions.
• Enables studying and referencing technical content directly within the coding environment.
• Supports glob patterns and folder inputs for flexible source aggregation.

Andrew Yng's aisuite provides a lightweight Python library to interact with various generative AI providers through a single, consistent API. The project also serves as the backend for OpenWorker, a desktop AI coworker that handles research and automation tasks. Users can integrate major cloud providers like OpenAI or run models locally via Ollama while keeping data on-premises.

• Abstraction layer simplifies switching between OpenAI, Anthropic, Google, and local Ollama models.
• Python-first design fits naturally into existing engineering workflows and scripts.
• Decouples application logic from specific vendor APIs to reduce lock-in risk.
• OpenWorker example demonstrates practical desktop automation using the same underlying interface.

The claude-video repository provides a plugin for Claude Code and compatible agents to process video content. It automates downloading, frame extraction, and transcription using yt-dlp and ffmpeg, then feeds the data to Claude for analysis. The tool supports global or project-scoped installation via npx with minimal initial configuration.

• Enables agentic video analysis by converting media to frames and transcripts for LLM consumption.
• Integrates with Claude Code, Codex, Cursor, and 50+ other agent skill hosts via npx.
• Automates dependency setup for yt-dlp and ffmpeg on first run across macOS, Linux, and Windows.
• Uses Whisper API only for videos without captions; otherwise, public captions are leveraged for free.