网页适配
adapt
by allens0104
Shortest English alias for web-adapt / universal-web-adaptation. Use this when adapting an unfamiliar public website generically before creating any site-specific rules.
安装
claude skill add --url https://github.com/openclaw/skills文档
Adapt skill
This is the shortest English alias for:
web-adaptuniversal-web-adaptation
It is best when you want to:
- adapt an unfamiliar site generically
- test search inputs, buttons, openers, and popups
- decide whether a dedicated site profile is actually necessary
Primary guidance
- try the generic path first
- tune runtime if the page is heavy
- follow popups or new tabs when needed
- only create a dedicated profile when the generic path is not robust enough
Quick invocation template
Use /adapt to adapt this unfamiliar site first, then tell me whether we should stay generic or create a dedicated site profile.
请用 /adapt 先通用适配这个网站，并告诉我是否真的需要专用 profile。
Pointer
For the full detailed playbook, see:
skills/universal-web-adaptation/SKILL.mdin this repository
相关 Skills
Claude API
by anthropic
Build, debug, and optimize Claude API / Anthropic SDK apps. Apps built with this skill should include prompt caching. Also handles migrating existing Claude API code between Claude model versions (4.5 → 4.6, 4.6 → 4.7, retired-model replacements). TRIGGER when: code imports `anthropic`/`@anthropic-ai/sdk`; user asks for the Claude API, Anthropic SDK, or Managed Agents; user adds/modifies/tunes a Claude feature (caching, thinking, compaction, tool use, batch, files, citations, memory) or model (Opus/Sonnet/Haiku) in a file; questions about prompt caching / cache hit rate in an Anthropic SDK project. SKIP: file imports `openai`/other-provider SDK, filename like `*-openai.py`/`*-generic.py`, provider-neutral code, general programming/ML.
并行代理
by axelhu
Use when facing 2 or more independent tasks that can be worked on without shared state - dispatches parallel subagents using sessions_spawn for concurrent investigation and execution, adapted for OpenClaw
高光制作器
by bwbernardweston18
>
相关资讯
研究团队利用开源小模型，在实验中构建了一款AI驱动的自适应计算机蠕虫，它能自主识别目标漏洞、自我复制并利用受害设备资源持续攻击。该蠕虫不依赖商业AI平台，成本几乎为零，标志着自主网络攻击从理论走向现实。
本文提出VLM-as-Teacher范式，利用视觉语言模型（VLM）作为教师，提取任务规则并构建可微分奖励，引导视频生成模型（VGM）通过测试时在线优化LoRA模块进行推理。在符号和通用视频推理基准上，该方法平均提升16.7个点，远超传统VLM-as-Solver范式（+0.4点）和Best-of-N采样（+2.2点），实现了高效且可泛化的视频推理。
本文提出未来前向动力学因果注意力（FFDC）机制，通过轻量级验证器联合推理预测动作、视觉动态和真实观测，判断预测轨迹是否仍可信，从而实现世界动作模型（WAM）的适应性执行。在RoboTwin基准上，该方法减少69.10%的模型推理次数和34.02%的执行时间，成功率提升2.54%；真实实验中成功率提升35%。该工作首次将动作块大小自适应问题建模为未来-现实一致性验证，在保持长程规划效率的同时提升复杂场景响应能力。