| 6.8 | clawhub | HTTP客户端 laosi-http-client provides a wrapper around urllib for sending GET/POST/PUT/DELETE/PATCH requests with custom headers, body serialization, auth support, timeout control, and retry logic. includes request history tracking and graphql helper. |
| 6.8 | clawhub | 文件整理 laosi-file-organizer automatically categorizes files in desktop and downloads folders by type (images, documents, archives, video, audio, code), generating a structured report with dry-run preview capability. |
| 6.8 | clawhub | API Tester Pro laosi-api-tester provides http request construction, response validation, and latency measurement for rest and graphql apis. includes batch testing, performance benchmarking, and test history persistence via json. |
| 6.2 | clawhub | Slack助手 laosi-slack wraps slack-sdk to send/search/manage channels and integrate notifications via webhook or socket mode. covers message posting, block kit formatting, channel ops, history search, and file upload with basic error recovery. |
| 5.8 | clawhub | 待办清单 laosi-quick-todo provides lightweight task management with voice/text input, priority sorting, completion tracking, and local json persistence. no external dependencies required. |
| 5.8 | clawhub | Database Helper Pro laosi-database-helper provides sql query building, schema management, and csv import/export for sqlite with python dataclass orm templates. covers basic crud operations and lacks multi-database support despite claiming mysql/postgresql compatibility. |
| 5.5 | clawhub | Cron表达式解析 laosi-cron-parser translates cron expressions to natural language, validates syntax, computes next execution times, and provides common presets. covers standard 5-field cron with basic step/range/list operators. |
| 5.3 | clawhub | JSON数据转换 laosi-json-transformer provides chainable operations for filtering, merging, grouping, and aggregating json data via a fluent api. covers basic transformations but lacks error handling and jmespath integration despite claiming it. |
| 5.2 | clawhub | 结构化头脑风暴 laosi-brainstorm structures ideation around four perspectives (constraints, alternatives, edge cases, priorities) with python dataclass implementation and markdown export. activates on brainstorm trigger phrases in chinese or english. |
| 5.2 | clawhub | Webhook Receiver Pro laosi-webhook-receiver provides a local http server for ingesting and logging webhook events with signature verification, event filtering, and handler registration. lacks production-grade error handling, rate limiting, and deployment guidance. |
| 5.2 | clawhub | 代码生成器 laosi-code-generator scaffolds projects and functions across python, javascript, and go with templated file generation and directory structure automation. provides three built-in templates and config file generation in yaml, json, toml formats. |
| 5.2 | clawhub | 剪贴板历史 laosi-clipboard-history maintains a 50-entry clipboard history for windows, automatically deduplicating and truncating to 200 chars per entry. search and recovery via python api, but lacks integration layer and production monitoring. |
| 4.9 | clawhub | Thunder Core thunder-core wraps the xunlei download client's com interface for windows, exposing task creation, commitment, and cancellation via python. covers basic download automation but lacks error handling depth and procedural rigor. |
| 4.9 | clawhub | 快速笔记 laosi-quick-note captures text or voice into tagged notes (idea/todo/readlater) with local json persistence and keyword search. auto-tagging relies on simple keyword matching and may misclassify ambiguous input. |
| 4.9 | clawhub | Skill Creator Enhanced skill-creator-plus provides templates and a publishing workflow for building skills on the opencode platform. it covers directory structure, validation rules, and clawhub integration but lacks concrete guidance on intent/input/output contracts and error recovery. |
| 4.8 | clawhub | 上下文记忆球 context-memory-ball introduces a conceptual framework for encapsulating conversation state into saveable units, enabling context switching and task parallelization. implementation details remain largely illustrative rather than actionable. |
| 4.8 | clawhub | 代码沙箱 laosi-code-sandbox outlines isolation strategies for executing untrusted code, including resource caps, network blocking, and pattern detection. the skill lacks actionable implementation steps and concrete runtime behavior specs. |
| 4.8 | clawhub | 多模型路由器 laosi-model-router classifies tasks by complexity and domain, then selects from a fixed model palette (claude, gpt, deepseek). the routing logic is pseudocode with decision tables but lacks concrete implementation details and error handling. |
| 4.5 | clawhub | 健康追踪 laosi-health logs water intake, sleep duration, and step counts into json files with basic statistics. lacks intent contracts, error handling, and procedural clarity on when/how to invoke each function. |
| 4.5 | clawhub | Systematic Debug systematic-debug outlines a hypothesis-driven debugging workflow (observe, hypothesize, isolate, verify, fix) with stated anti-patterns, but lacks decision logic, outcome signals, and concrete failure mode guidance. |
| 4.3 | clawhub | Blog Watcher blog-watcher monitors website and rss feed updates via dual detection (hash comparison and feed parsing), outputs change summaries. v2 adds diff comparison and exclusion rules, though implementation details are absent. |
| 4.3 | clawhub | Audio Transcriber audio-transcriber wraps openai whisper for speech-to-text conversion with keyword-based emotion detection and basic audio signal analysis. covers transcription, translation routing, and noise reduction across multiple subcommands. |
| 4.2 | clawhub | 迭代优化器 laosi-iteration-opt proposes a multi-round code evaluation framework with weighted scoring across correctness, performance, readability, and maintainability. lacks concrete procedural steps and operational validation. |
| 4.2 | clawhub | Windows剪贴板管理器 laosi-clipboard provides code snippets for windows clipboard operations including history tracking, multi-slot storage, and text formatting. lacks executable intent definitions and error handling. |
| 4.2 | clawhub | Code Review Pro laosi-code-review applies a four-perspective framework (security, performance, correctness, style) to identify code issues with severity ratings and fix suggestions, outputting pass/block verdicts. implementation is partial and pattern-matching based rather than rigorous static analysis. |
| 4.2 | clawhub | Spotify控制 laosi-spotify wraps spicetify-cli to control playback, search tracks, and manage playlists via subprocess calls. lacks error handling, input validation, and connection recovery. |
| 4.2 | clawhub | 会话日志分析 laosi-session-logs provides keyword search and time-filtered retrieval of conversation history from local log files. lacks error handling, output validation, and clear invocation contract. |
| 4.2 | clawhub | Obsidian笔记助手 laosi-obsidian provides local file-based note operations (create, search, tag, backlink detection, graph export) via python scripts that directly manipulate markdown files in an obsidian vault without api integration or error handling. |
| 4.2 | clawhub | Collaborative Agent collaborative-agent implements multi-role task orchestration with planner, executor, reviewer, and summarizer personas. lacks concrete failure handling and assumes external capability resolution. |
| 4.2 | clawhub | Chuwi Minibook chuwi-minibook collects system metrics (cpu, memory, disk, battery, temperature) from minibook devices via psutil with three invocation modes: full report, quick overview, and battery status only. |
| 4.1 | clawhub | Laosi Mcp Security Audit laosi-mcp-security-audit claims enterprise-grade vulnerability scanning and malware detection for mcp servers but lacks concrete procedural steps, specific invocation patterns, or failure mode handling. |
| 3.9 | clawhub | Ai Chat Enhancer ai-chat-enhancer manages conversation history, token counting, response caching, and prompt templating for llm interactions. the skill lacks actionable procedures and edge case handling. |
| 3.2 | clawhub | Product Analysis Skill product-analysis-skill claims market analysis and competitor research capabilities but provides no documented intent, input schema, procedural steps, decision logic, or output contract to verify what it actually does. |
| 3.2 | clawhub | Continuous Learner continuous-learner claims to orchestrate a closed-loop pipeline across insight collection, knowledge ingestion, trend extraction, and skill generation, but lacks actionable procedure detail and concrete failure handling. |
| 3.2 | clawhub | Auto Insight Engine auto-insight-engine aggregates content from six sources (hackernews, github, wikipedia, rss, arxiv, web search) and applies nvidia nim analysis to trending keywords, but the skill.md conflates unrelated bilibili video tutorials with the core engine logic. |
| 2.3 | clawhub | Transformer Optimize E72719 transformer-optimize-e72719 aggregates video-sourced material on building transformer models for multivariate time series forecasting in matlab, with minimal procedural detail and confused scope across unrelated topics. |
| 2.3 | clawhub | Super Search Dee798 super-search-dee798 claims search functions outperform find functions by claiming 100x superiority without defining either function, demonstrating the claim, or providing actionable steps. |
| 2.3 | clawhub | Database Helper 99a323 database-helper-99a323 aggregates video-sourced notes on database tools and sun-panel configuration, lacking coherent procedure and actionable guidance for practical deployment or troubleshooting. |
| 2.3 | clawhub | Skill Learner 818d6a skill-learner-818d6a aggregates video-sourced tutorials on hermes skill bundles and opencode patterns, curating bilibili references without executable procedures or outcome specifications. |
| 2.3 | clawhub | Add Flashcard add-flashcard wraps a spaced repetition system via capability_executor but lacks procedural detail, input/output contracts, and failure handling. the document is primarily metadata and bilibili video references. |
| 2.3 | clawhub | 模型微调 9acd6f 9acd6f attempts to document qwen2.5-7b model fine-tuning based on video content, but lacks actionable steps, actual implementation details, and coherent structure beyond aspirational headers. |
| 2.3 | clawhub | 代码生成 876ae0 876ae0 aggregates video-sourced fragments about code generation tools and simulink workflows, but lacks coherent procedural content, actionable inputs, or verifiable outputs. |
| 2.3 | clawhub | Auto Agent 4974 auto-agent-4974 aggregates bilibili video references on ai agent development without providing actionable procedural steps, concrete examples, or decision logic for building agents. |
| 2.3 | clawhub | Ai Summary ai-summary claims to save ai knowledge summaries to a knowledge base but provides no actionable procedure, error handling, or clear contract between input and stored output. |
| 2.3 | clawhub | Comfyui 7e6098 comfyui-7e6098 collects video references on comfyui setup and ai image generation, primarily targeting intel gpu configuration, but lacks actionable procedural depth. |
| 2.3 | clawhub | Data Analyzer From Bilibili skill purports to analyze data via statistical description, anomaly detection, and trend analysis, but lacks actionable procedure, decision logic, or output contract. appears to be auto-generated bilibili video references without executable steps. |
| 2.3 | clawhub | Deepseek Reasonix F074f8 deepseek-reasonix-f074f8 aggregates bilibili video content about deepseek ai agents and ppt generation, but lacks actionable steps, technical specifics, or decision logic. |
| 2.3 | clawhub | 手机摄影 3e7019 3e7019 aggregates mobile photography tutorials from chinese video platforms without substantive procedural detail. skill lacks actionable steps, decision logic, and failure recovery paths. |
| 2.2 | clawhub | 数据分析 6450d8 6450d8 aggregates video-sourced data analysis material covering spss workflows, response surface analysis, and design optimization, but lacks actionable procedure depth and repeats content across merged sections. |
| 2.1 | clawhub | Text Diff Be5bd0 text-diff-be5bd0 purports to teach react diffing algorithm concepts via video, but contains no actionable procedure, no decision logic, no output contract, and references unrelated bilibili links. |
| 2.1 | clawhub | Swe Bench A6e0fc swe-bench-a6e0fc conflates ai code evaluation benchmarks with aspice compliance templates across repeated, near-identical sections. lacks actionable procedure, decision logic, or outcome measurement. |
| 2.1 | clawhub | Ai Summary B037a3 ai-summary-b037a3 documents video-sourced instruction on summary writing without defining concrete outputs, actionable steps, or success criteria. |
| 2.1 | clawhub | Deeptutor F75dc7 deeptutor-f75dc7 documents local deployment versus desktop variants, sourced from a video tutorial. actual procedures and decision logic are absent, leaving only abstract concepts and external references. |
| 2.1 | clawhub | Devin 192ac1 devin-192ac1 is a corrupted skill document that conflates an ai coding platform with painting tutorials and basketball highlights, lacking coherent intent or actionable procedures. |
| 2.1 | clawhub | Deep Search 244e83 deep-search-244e83 claims to teach a comparison between search and find functions via video, but provides no actionable steps, code examples, or concrete guidance on implementing or applying the technique. |
| 2.1 | clawhub | Ai测试 09ed64 ai-09ed64 purports to cover ai testing fundamentals and api gateway recommendations, but consists primarily of duplicated video comments and promotional copy without actionable procedures, decision logic, or success criteria. |
| 2.1 | clawhub | Speak 69378f speak-69378f aggregates video-sourced content about the speak product and speaknow media, combining unrelated bilibili references without actionable procedures or clear intent definition. |
| 2.1 | clawhub | Academic Research Skills academic-research-skills references ai workflows for research stages (topic selection, outlining, writing, review, response) but lacks concrete procedural definition, actionable steps, or integration details. |
| 2.1 | clawhub | Add Flashcard 4c0314 add-flashcard-4c0314 references a bilibili video tutorial on building a geography flashcard app with image support, but provides no actual implementation guidance, code samples, or actionable steps. |
| 2.1 | clawhub | 视频剪辑 69cfc9 69cfc9 aggregates bilibili video snippets about video editing software selection and techniques, but lacks actionable guidance, concrete software recommendations, or decision frameworks. |
| 2.1 | clawhub | 整整600集 E1aeca 600-e1aeca aggregates bilibili video references for python web scraping instruction, lacking concrete steps, decision logic, failure handling, or measurable outcomes. |
| 2.1 | clawhub | 文生视频 16ea34 16ea34 documents text-to-video workflows using comfyui, sourced from bilibili tutorials. content is heavily duplicated, lacks actionable steps, and mixes unrelated references (game footage, door sales). unsuitable for practical deployment. |
| 2.1 | clawhub | 多智能体 3c4b57 3c4b57 aggregates video-sourced multi-agent notes from bilibili educators covering langgraph, coze, and llm council patterns, but lacks executable procedures, failure handling, or concrete implementation details. |
| 2.1 | clawhub | Ai Video From Bilibili ai-video-from-bilibili claims to save ai video generation summaries to a knowledge base, but lacks actionable procedure, concrete outputs, or failure handling. mostly bilibili video links and metadata noise. |
| 2.1 | clawhub | Ai助理 Bb6cad ai-bb6cad aggregates fragmented video content about ai assistants, cad tools, and gaming guides with no coherent procedure, decision logic, or output contract. the skill lacks actionable structure. |
| 2.1 | clawhub | Api Tester From Bilibili api-tester-from-bilibili is an auto-crystallized skill with corrupted metadata. the document conflates unrelated bilibili videos (train schedules, laptop teardowns, headphones) with api testing concepts, lacking coherent procedure or actionable steps. |
| 2.1 | clawhub | Ai安全 B9b1fb skill document claims to teach ai security and penetration testing but contains only vague platitudes, repetitive boilerplate, suspicious urls, and no actionable technical content. |
| 2.1 | clawhub | Deepseek Reasonix deepseek-reasonix is a deepseek terminal programming agent intended to reduce long-session costs, but the skill.md contains only bilibili video references and fragmented metadata with no actionable procedure or clear intent definition. |
| 2.1 | clawhub | 低代码Ai 7bbc51 ai-7bbc51 aggregates bilibili video clips on english language learning and ai topics with no actionable low-code workflow, repeated sections, and conflated content sources. |
| 1.9 | clawhub | Ai编程 0b1387 ai-0b1387 is a corrupted skill artifact containing duplicated sections, off-topic content (bus routes, water heaters, train stations), and generic placeholder steps with no actionable substance or technical depth. |
| — | clawhub | Ai Chat Enhancer 0f5a92 从抖音视频学习: ai-chat-enhancer — Windows部署教程，使用Chat2API零成本接入主流模型（如GLM、Kimi、Qwen等）并共享给朋友、同事 |
| — | clawhub | Api Tester Dd7739 从抖音视频学习: api-tester — 🔄 New API进阶教程（下）| 对接各类AI渠道完整攻略，轻松接入第三方平台 |
| — | clawhub | Auto Agent 4974 F58bc0 从抖音视频学习: auto_agent_4974 — 一口气搞懂 Notion Custom Agent（测评与教程） |
| — | clawhub | Audio Device Diagnostic 音频设备诊断技能 — 列出系统所有可用音频输入设备，检测设备是否正常工作。 Trigger: 当音频设备不工作、需要查看可用麦克风、或选择正确的音频输入设备时。 |
| — | clawhub | Code Sandbox 代码沙箱 - 原创技能。安全执行未验证的AI生成代码，防止恶意代码、系统破坏或意外损害。适用于代码审查、安全验证、AI编程辅助等场景。 |
| — | clawhub | 开源模型 13b59f 从抖音视频学习: 开源模型 — 手把手教你选择开源AI模型！一键分析哪些模型下载量最高，热度最高！ |
| — | clawhub | 办公软件 1896ae 从抖音视频学习: 办公软件 — 原来是办公软件啊妈妈从现在起电脑上只用办公软件我行吧 |
| — | clawhub | Clipboard History 剪贴板历史管理 — 自动保存复制内容，随时找回 |
| — | clawhub | 视频调色 705d12 从抖音视频学习: 视频调色 — 用Pocket3把你的生活变成电影——Pocket3调色教程零基础到进阶 |
| — | clawhub | 代码审查 89d9f5 从抖音视频学习: 代码审查 — OpenAI Codex教程 #3 - 代码审查 |
| — | clawhub | Academic Research Skills 606a56 从抖音视频学习: academic-research-skills — Academic Research Skills 完整实操教程——13条AI指令覆盖选题/提纲/撰写/审稿/回复全流程 |
| — | clawhub | Kl8 Data Query China Happy 8 (KL8) lottery historical data — latest draw, specific issue, recent draws, and number frequency analysis. |
| — | clawhub | Model Center Unified interface to 42+ NVIDIA NIM API models — LLM chat, vision, embeddings, image generation, with price comparison and model recommendation. |
| — | clawhub | Self Improvement Daemon 自我改进守护进程，持续学习优化系统 |
| — | clawhub | Self Improvement Daemon 2ed609 从抖音视频学习: self-improvement-daemon — 如何利用DTL (Daemon Tools Lite)来安装iso类游戏？（以老游戏安装为例） |
| — | clawhub | Side Hustle Search for side hustle opportunities |
| — | clawhub | Skill Validator 4f8f6e 从抖音视频学习: skill-validator — 用魔法创造魔法，如何使用 skill-creater创建skill |
| — | clawhub | Solve Problem Search knowledge base and research solutions |
| — | clawhub | Sora 478958 从抖音视频学习: Sora — Sora2 保姆级教程：从注册到出片，全网最详细教程，完全免费！ |
| — | clawhub | Agent开发 09770e 从抖音视频学习: Agent开发 — 【Agent】全B站最全最细的Agent开发教程，2026最新！全程干货，允许白嫖！从入门到企业级实战掌握Agent智能 |
| — | clawhub | Agent Builder 5bce38 从抖音视频学习: agent_builder — ☆新发售☆【辅助工具DLC】RPG MAKER MZ／MV-Tileset Builder（像素美术编辑工具） |
| — | clawhub | Ai工作流 43de14 从抖音视频学习: AI工作流 — 2026Coze 扣子工作流智能体保姆级搭建教程，专为零基础 AI 新手量身打造，10 分钟无代码手把手实操，解决不会设 |
| — | clawhub | Ai Agent 441222 从抖音视频学习: AI Agent — 2026吃透AI Agent智能体全套系统教程，手把手教你从0到1打造企业级AI Agent智能体，学完即可就业！拿走不 |
| — | clawhub | Data Analyzer 622b06 从抖音视频学习: data-analyzer — 【化工设计】换热网络需要完成什么？初识Aspen energy analyzer |
| — | clawhub | Collaborative Agent 1fe479 从抖音视频学习: collaborative-agent — AI给自己写工具？字节团队出品 MUSE-Autoskill 让Agent实现技能自进化，推动skills成为可复用的知 |
| — | clawhub | Ai绘画 Ddcd9b 从抖音视频学习: AI绘画 — 【全108集】目前B站最全的AI绘画网课SD+ComfyUI+Midjourney保姆级教程，2025最新版！包含所有干 |
| — | clawhub | Ai Video 55eeb0 从抖音视频学习: ai-video — 【高清MV】洋澜一 - 菩萨菩萨你睁睁眼【Pinyin Eng Sub AI音乐视频 繁体拼音英文动态字幕 】 Yang |
| — | clawhub | Database Helper Auto-crystallized skill from successful execution |
| — | clawhub | Team Collaboration Ca6749 从抖音视频学习: team-collaboration — 【人均8美元出头即薅到48个月GPT Team会员教程】置顶评论分享优惠码，每月17.4美元畅用GPT Business |
| — | clawhub | Think Expand 想核心（大脑主核心），统筹听/看/说/做四大核心，集成MemPalace记忆检索、自我进化、闭环学习，事件驱动决策，永动机式持续运行，仅OpenCode内部使用。 |
| — | clawhub | Translation Assistant Translation assistant capability |
| — | clawhub | 技能创建指南 技能创建指南 - 创建/发布ClawHub技能的完整指南，含模板、最佳实践和发布流程 |
| — | clawhub | GitHub操作助手 GitHub操作助手 - PR/Issue/代码搜索/仓库管理，基于gh CLI。v2新增：CI状态监控、Release管理、Code Review辅助 |
| — | clawhub | 阅读助手 阅读助手 - 贴链接或粘贴文字，自动提炼4点：主旨/论据/行动/疑问，不用读全文。支持网页抓取和本地文本 |
| — | clawhub | 快速草稿 快速草稿 - 说几个要点自动生成结构化邮件/消息/周报草稿，内置多种模板(followup/thankyou/request/report) |
| — | clawhub | 习惯追踪 习惯追踪 - 每天打卡，自动计算连续天数/历史最佳/总打卡数，多习惯管理，本地JSON持久化 |
| — | clawhub | 系统化调试 系统化调试 - 假设驱动调试：观察→假设→隔离→验证→修复。附带Binary Search调试树、Minimal Reproducer生成 |
| — | clawhub | 实现计划生成 实现计划生成 - 从需求生成多阶段实施计划：基础→逻辑→接口→测试→部署。含模板、依赖分析、验收标准 |
| — | clawhub | 测试驱动开发 测试驱动开发 - 严格红/绿/重构循环，一次一个垂直切片，测试先行。含pytest/mock示例和覆盖率检查 |
| — | clawhub | 文本对比工具 文本对比 - 对比两段文本差异，逐行diff，高亮变更，支持side-by-side和unified格式输出 |
| — | clawhub | Email Helper Pro 邮件助手 - 编写/发送/管理邮件，支持SMTP发信和IMAP收信，多模板草稿，自动归档 |
| — | clawhub | Scheduler Pro 任务调度 - 定时任务/周期性任务/一次性提醒，cron语法支持，持久化存储，下次执行时间自动计算 |
| — | clawhub | Data Analyzer Pro 数据分析 - 加载CSV/JSON自动计算统计描述(均�?中位�?标准�?极�?，异常检测，趋势分析，结果本地持久化 |
| — | clawhub | Git Commit Helper Pro Git助手 - 自动生成提交信息/查看状�?分支管理/diff分析，基于Conventional Commits规范 |
| — | clawhub | PDF Processor Pro PDF处理 - 文本提取/表格识别/文档合并/页面拆分/元数据读取，纯Python实现 |
| — | clawhub | Crypto Encoder Pro 编码工具�?- Base64编码/解码、URL编码/解码、SHA256/MD5哈希计算、UUID生成、JWT解析 |
| — | clawhub | Contract Review Skill AI-powered contract review that identifies risky clauses, missing provisions, and compliance issues in legal documents for informed decision-making. |
| — | clawhub | Karpathy编程四大原则 AI编程四大原则 - 源自Karpathy法则 (forrestchang/andrej-karpathy-skills 94.2k⭐)。用于在AI编程时强制执行先思考、保持简单、精准修改、目标驱动四大原则。适用于AI代码审查、代码生成、修复bug等场景。 |
| — | clawhub | MemPalace记忆系统助手 MemPalace记忆系统助手 - 基于老四的AI记忆系统。提供结构化记忆存储、快速检索、记忆组织等功能。适用于需要持久化知识、跨会话记忆、语义搜索等场景。 |
| — | clawhub | 工作流验证器 工作流验证器 - 在执行前验证工作流的正确性、安全性和完整性。基于Karpathy法则，强调先思考、保持简单、目标驱动的原则。适用于代码审查、工作流执行、任务规划等场景。 |
| — | clawhub | 熵管理系统 熵管理系统 - 原创技能。用于管理和控制AI会话中的熵（无序度），包括上下文精简、状态重置、注意力聚焦等功能。适用于长时间会话、复杂任务、多步骤工作流等场景。 |
| — | clawhub | 意图分类器 意图分类器 - 原创技能。实时分析用户输入，识别真实意图并路由到对应处理流程。支持代码、提问、请求、闲聊等类型识别，适用于AI助手任务调度、智能路由、对话理解等场景。 |
| — | clawhub | 幻觉检测器 幻觉检测器 - 原创技能。检测AI生成代码中的幻觉内容，包括不存在的API、错误的参数、虚假引用等。适用于代码审查、AI编程辅助、质量保证等场景。 |
| — | clawhub | 上下文窗口优化器 上下文窗口优化器 - 原创技能。智能管理AI上下文窗口，自动压缩历史、去除冗余、保留关键信息，节省token提升效率。适用于长会话、多文件、大项目等场景。 |
| — | clawhub | 需求澄清器 需求澄清器 - 原创技能。自动追问模糊需求，减少返工和误解。适用于AI编程、需求分析、任务交接等场景。 |
| — | clawhub | Windows应用控制器 Windows应用控制器 - 原创技能。让AI通过自动化技术控制Windows应用程序，包括打开/关闭应用、点击按钮、填写表单、截取屏幕等操作。适用于GUI自动化、测试、数据录入等场景。 |
| — | clawhub | Windows文件管理器 Windows文件管理器 - 原创技能。让AI执行Windows文件操作，包括文件/文件夹的创建、复制、移动、删除、搜索、重命名等。适用于文件整理、批量处理、自动化工作流等场景。 |
| — | clawhub | RSS监控 RSS监控技能 - 监控RSS/Atom订阅源，检测更新，获取新内容。 |
| — | clawhub | 博客监控 博客监控技能 - 监控博客和网站更新，通过RSS或网页变化检测。 |
| — | clawhub | Notion助手 Notion API助手 - 创建和管理Notion页面、数据库，支持笔记和知识管理。 |
| — | clawhub | 自动学习技能 自动学习技能 - 从对话中自动提取知识并创建技能。基于Memento-Skills的Read-Write循环机制。适用于AI Agent自进化、知识积累、技能自动创建等场景。 |
