SKILLS_INDEX

A Claude Code skill that makes AI agents communicate in terse caveman speech, cutting ~75% of output tokens while maintaining full technical accuracy. Offers multiple intensity levels (Lite/Full/Ultra/Wenyan) for different compression needs, plus utilities for terse commits, one-line code reviews, and input token compression. Based on research showing brevity constraints can improve model accuracy by 26 percentage points.

Make websites accessible for AI agents. Automate tasks online with ease. A powerful AI browser agent that enables browser automation for websites. Supports multi-LLM integration (OpenAI, Anthropic, Google), cloud hosting with stealth and proxy rotation, self-hosted option, CLI tools, custom tool extensions, and MCP integration. Common use cases include form filling, online shopping automation, and PC part searches.

Plan and optimize website hierarchy, navigation, URL structure, and internal linking to improve SEO and user experience.

Ready-to-use AI agent skills for research, science, engineering, and finance. Features the ESM toolkit for state-of-the-art protein language modeling, supporting generative protein design (ESM3), structure prediction, inverse folding, and embeddings (ESM C) with local and cloud inference.

Persistent context and memory plugin for AI agents across sessions. Captures, compresses, and injects relevant session history into future interactions. Supports OpenClaw, Claude Code, Codex, Gemini, and more.

A minimal browser automation harness built on Chrome DevTools Protocol that grants LLMs complete freedom to execute any browser task, featuring a self-healing mechanism for dynamic code writing and fixing mid-execution.