maref
We audited 100 open-source Agent projects – 73% have permission overreach
开源项目、精巧工具,还有让人眼前一亮的点子。
从 Hacker News 社区讨论中,找到你的下一份灵感。
We audited 100 open-source Agent projects – 73% have permission overreach
Omegacode: Code based orchestration for any coding agent
Paddock – Ranks and runs LLMs on your Mac from live GGUF headers
Moraine: Unified Agent Tracing
Teamclaude rewrite in rust, run multiple Claude accounts in parllel
An AI gateway that signs a receipt for every LLM response
Clodex – Open-source agentic IDE with governed execution and graph memory
Aict – Unix coreutils that output XML/JSON, built for AI agents
A model-routing benchmark – the routers optimize the wrong axis
Claudeisms: A list of words Claude Code overuses
Two-tier-memory – queryable long-term memory for AI coding agents
Skills as a Service via MCP –> Coding Agent Skill Library
Honeyprompt: LLM-first honeypot that talks back
Pomi – standalone voice agent running on the Rabbit R1
Betlang – A programming language detection model in 50kb
Looperators – work with agent loops visually
LangChain's OpenWiki agent based on PI.dev harness
Chrome Extension Claude Token Usage Bar and Context Use for Claude.ai