Trending
skills & tools · github stars- Gathin23/cleanyourtv
Debloat any Android or Google TV over Wi-Fi and get an Apple TV-style home.
- simbasjungles/free-my-tv
A fast, ad-free home screen for Google TV, LG webOS and Samsung Tizen. One button to undo.
- t3dotgg/tstack
The agent skills I use every day. Works with Claude Code and Codex.
- andyhuo520/digital-human-koubo
AI digital-human talking-head videos from one photo: Renoise MCP generation + 5 editing styles + 48 creator-grade packaging components. Agent skill for Claude Code / Codex / OpenClaw / WorkBuddy …
- DenisSergeevitch/ex-ussr-kinship-research
Evidence-first genealogy skill for AI agents (Codex, Claude Code): church books, civil, military, deportation and repression records across the former USSR. Replies in your language.
- morganlinton/BeautifulRust
Agent Skills that teach coding agents to write, review, and port idiomatic Rust, plus a tested clippy config. Works with Claude Code, Codex, Copilot, Cursor, and Gemini CLI.
Tools & Packages
- agent-usage-guard – a Claude Code mod to protect your usage window
A Claude Code mod to cap usage windows—practical utility for controlling API costs and session budgets.
Patterns & Techniques
- Claude Mods: Flipping the Harness Inside Out
Introduces Claude Mods as an 'Attention Interface' that inverts the harness paradigm to prioritize human-facing interaction over machine-facing tool orchestration.
- Anthropic’s new guide (Oct 22) argues most agents should have fewer subagents: one main loop, skills and tool search instead. Includes “we cut Claude Code’s system prompt by over 80%”
Anthropic官方指南建议默认使用单agent循环加技能/工具搜索，减少子agent，并分享Claude Code精简系统提示80%的经验。
- I replaced my Samsung TV’s home screen with Claude Code
展示用Claude Code构建三星电视启动器，通过模拟TV API快速迭代，替换广告屏。
- I got tired of checking whether Claude Code needs me. So I gave it a physical status light.
利用Claude Code hooks连接物理灯，实时显示agent是否等待用户输入，提升协作体验。
Discussion
- Claude Code slyly adding users' emails to outbound HTTP
Reports that Claude Code leaks user email in outbound HTTP headers—critical privacy/security concern for harness operators.
- Donating Agent Skills to the Agentic AI Foundation
Anthropic donates Agent Skills standard to the Agentic AI Foundation under Linux Foundation, ensuring open, vendor-neutral governance—key for MCP and skill portability.
- No coding or gamedev background, 6 weeks with Opus 5.5 + Godot — from drawings to a playable game
无编程背景用户在Claude Code（Opus 5.5）和Godot帮助下六周内创作可玩游戏，分享开发流程。
Research
- Navigating Epistemic Parity in LLM Agents: A Benchmark for Cross-Source Conflict Resolution Between Memory and Procedural Skills
Introduces a benchmark and taxonomy for how LLM agents resolve conflicting instructions between episodic memory and procedural skill files like SKILL.md, a core issue for agent harnesses.
- Can Agentic Code Review Catch Vulnerabilities Before They Are Merged?
Introduces SecPRBench, a benchmark for evaluating agentic security review on real-world bug-introducing commits, providing a realistic test for pre-merge vulnerability detection by coding agents.
- How Do Small Coding Agents Use Code Graphs and Embeddings?
Ablation study of small coding agents using embedding search and code-graph tools on SWE-bench Verified, stratified by hop depth, yielding actionable findings on when and how repo-level structure aids agents.
- A Multi-Domain Study on the Effectiveness of LLMs at Writing Property-Based Tests
Studies LLM-based property-based test generation across domains, showing that tools like Proptest-AI and Agentic-PBT falter outside Python; directly relevant to agent capabilities in testing.