Tag
作者开源了 agent-engineer 插件，适用于 Claude Code、Codex CLI、Cursor 等，包含 11 个技能、4 个子代理和 5 个钩子；盲测 A/B 显示 Claude Code (Opus 5.5) 使用插件后任务需求满足率从 34% 提升到 53%。
A 20-year software engineer compares Gemma4-31B and Qwen3.8-27B (Q4, via Codex CLI and OpenCode CLI) on codebase analysis and greenfield C#/WPF app authoring, finding both workable locally but clearly behind GPT 6.1-Sol, which delivered a one-shot A+ result.
AKM 是一个本地优先的开源 CLI，帮助团队在 Claude Code、Cursor、OpenCode 等多个 shell 能力编码助手之间共享一个可复用的技能、工作流和代理知识库，支持就地索引、聚焦检索和可分享的打包捆绑。
发布了一个包含28个可移植Agent Skills的开源目录，为Claude Code、GitHub Copilot、Codex和Cursor等编码助手提供结构化推理框架，涵盖决策、诊断、系统思考、风险、战略与创新等场景。
Codync launches as an open-source, MIT-licensed alternative to Grok Bot, Muse, and Dots, letting users run Claude Code, Codex, Cursor and 40+ agents as named bots on their own machine, with remote approval from iPhone, Mac, Linux or terminal via end-to-end encrypted channels.
OpenHands is exploring the future of agentic software development through specialized agents working across the SDLC alongside existing developer tools, with an emphasis on openness and cost efficiency as agent usage scales.
xAI's Eric Zakariasson announced new engineering bots in the Grok Bot Marketplace, highlighting Grok Bot's enhanced coding capabilities including handing off tasks to Cursor, managing PRs via GitHub and Origin plugins, and sharing video demos of built software.
A tweet hyping the upcoming Fable model while quoting Cursor's announcement that Sonnet 5.5 is now available and performs on par with Opus in many tasks.
The author critiques Claude's new Projects beta for limited cross-session memory and describes using Vilix AI's shared memory to integrate tools like Claude, Codex, and Cursor for a seamless workflow.
LangChain's Managed Deep Agents now function as MCP servers, allowing coding assistants and other agents to delegate tasks like research and analysis via a standard protocol endpoint.
Qodo has launched the Agentic Toolbox, enabling Claude Code, Codex, Kiro, and Cursor to access its review engine, codebase knowledge, and team rules.
Early access to new Xiaomi MiMo-X-Pro-Preview and MiMo-X-Flash-Preview AI models for professional workflows, focusing on complex reasoning and multi-agent collaboration.
Inception Labs releases Mercury 2.5, a diffusion-based language model with improved intelligence, speed, and cost-efficiency for production use in search, voice, and coding.
The author discusses their experience with AI models like GLM 5.3 Flash and DeepSeek V4 Flash on Open Router, finding them unreliable compared to Claude or Devin, and seeks a trustworthy agentic harness for effective model handling.
The tweet comments on Fable 5.1's impressive benchmarks in agentic research and coding but views it as incremental, expressing more interest in OpenAI's Astra for its potential persistent memory.
The author notes that after a year of AI coding harness adoption, developers often use multiple tools in companies, but consolidation is expected due to consistency, cost, and compliance reasons.
This tweet announces a trending GitHub repository by Matt Pocock that provides composable agent skills for real engineers to use with coding agents like Claude Code and Codex, aimed at fixing common failure modes in AI-assisted development.
The author describes their experience migrating from Claude Code to Codex, claiming that Anthropic's products have no stickiness and can be easily replaced.
Google introduces Custom Agents in Antigravity 2.0 and CLI, a feature that enables specialized AI agents to improve coding productivity by reducing context bloat and allowing division of labor.
A tweet highlights the promise of model routers and 'model melding', referencing the launch of Not Diamond Code, an intelligent model router for coding agents that cuts costs by 20-65%.