$ loading_
35211 Skills & MCP tools — copy a prompt, zero pre-install, into Claude Code / Codex / Cursor.
🎓 Skill = teach the AI a way to do a job (make slides / review code / write reports), copy the prompt and go — no account needed.
Write, understand, and debug code in the Lobster programming language.
Read, edit, and write Feishu documents for faster cloud collaboration.
Verify an OpenClaw release is fully published and working across all channels.
Convert text to speech locally and offline with sherpa-onnx, no cloud needed.
Create, refine, validate, and restructure AgentSkills and SKILL.md files.
Create, organize, search, and export Apple Notes from the macOS command line.
Manage Feishu document and file sharing, access permissions, and collaborators.
Fetch GitHub issues, create fixes, open PRs, and handle reviews.
Debug Node.js apps with inspect, breakpoints, heap, and CPU profiling.
Summarize CodexBar local cost logs by model for Codex or Claude.
Manage Notion pages, content, data, and workflows via CLI and API.
Manage files, folders, and storage in Feishu cloud drive.
Build quick throwaway prototypes to validate options and deliver a verdict.
Triage inbox messages, route intents, track replies, and generate follow-up summaries.
Generate concept, architecture, flow, and whiteboard diagrams in multiple formats.
Review, triage, and manage OpenClaw issues and pull requests faster.
Transcribe audio into text or speaker-separated transcripts with OpenAI APIs.
Run, debug, monitor, and summarize OpenClaw release CI workflows.
Helps maintainers triage, clean up, and resolve OpenClaw secret scanning alerts.
Analyze and optimize OpenClaw test performance, memory use, and coverage slowdowns.
Investigate test memory growth, heap leaks, and OOM root causes.
Discover and control Sonos speakers for playback, volume, status, and grouping.
Automate web page workflows, login checks, tab handling, and recovery steps.
Preview local channel message flow fixtures to inspect conversation flow behavior.
从访谈内容留存、转写回看，到主题归纳与产品建议，一套串起用户研究整理流程。
从流量采集、营销归因到数仓查询与看板输出，一套串起自动取数和分析闭环。
从读取设计稿、提取上下文到映射组件库与同步评审，这套组合很适合设计到前端落地的协作链路。
In-depth guides on Skill / MCP setup and security — beyond the directory listings
普通聊天 AI 很难真正读懂大型仓库，因为它缺少结构、上下文和可验证的检索路径。本文从 0 到 1 搭一个只面向本地代码库的 Cursor 助手：能查结构、追调用链、看 diff、给重构建议，还能尽量压住幻觉。
Agent-hop 是一个基于 Rust 的 TUI，支持在进行中的 Claude Code 聊天里切换到其他编程代理，尽量不丢失上下文。项目作者提到，它最早源自一个简历与搜索工具，后来因为使用过程中的阻碍，单独做成了现在的界面。
Harden.run 表示，针对能够生成任意代码的 Coding Agent，单靠传统红队测试并不够。他们通过后训练网络安全小模型、调整推理方式，并结合 inline reference monitoring 等程序分析手段，在 LinuxArena 和 SleightBench 上取得了超过 GPT5.5-xhigh 的成绩。