菜单
Search the web using Kagi and fetch page contents. Use when you need current information, documentation, or to look up APIs/libraries.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | websearch |
| description | Search the web using Kagi and fetch page contents. Use when you need current information, documentation, or to look up APIs/libraries. |
Search the web using Kagi and fetch/extract page contents.
kagi-search "your query"
Returns search results with titles, URLs, and snippets.
kagi-search --json "your query"
Returns JSON with:
quick_answer.markdown — AI-generated summaryresults[].title, results[].url, results[].snippetkagi-search --json --links 10 "your query"
curl -sL "https://example.com" | head -500
For cleaner extraction from HTML, pipe through a text extractor or use curl with appropriate flags.
kagi-search --json "nix flake lock update single input" — get overview + linkscurl -sL "<url>" — get full page content from a promising resultGet the user's attention by speaking a short message via TTS and posting a desktop notification. Use when a long-running task finishes, when you need input to continue, or when something needs human attention.
Run long-running commands in a tmux session and capture their output. Use when you need to start a build/test/server that takes minutes, run multiple commands in parallel, or interact with a process across tool calls.
Reverse engineer and debug binaries using Ghidra (static analysis, decompilation) and GDB (live debugging of running processes). Use when the user wants to analyze binaries, decompile functions, debug running processes, set breakpoints, or inspect memory.
Manage HedgeDoc pads on pad.lassul.us — create, read, edit, and get info on collaborative markdown documents. Use when the user wants to work with pads, notes, or shared documents.
Control Firefox browser from the command line. Use for web automation, scraping, testing, or any browser interaction tasks.