菜单
Visit a webpage and extract its content as markdown, or fetch images. Use for reading articles, documentation, or any web page content. Handles both HTML pages (via Jina Reader) and image URLs (downloads and saves locally).
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
Structured planning workflow for substantive coding work. Use at the start of every new major feature, non-trivial bug fix, refactor, or implementation request. Do not use this skill for trivial work, tweaks, or quick-win changes.
Store evidence for retrieved or recalled factual claims about external reality before stating them. Triggers — snippets from web-search/visit-webpage, statements about real entities you did not observe in this session (people's roles, company facts, dates, prices, versions, statistics, current status), quotes or paraphrases of others' words, and recalled facts from documentation. Do NOT use for descriptions of work you just performed (code changes, file edits, command output, anything visible in a diff or artifact), your own reasoning, recommendations, or design opinions; stable definitional knowledge ("Python is a language"); the user's own statements; or hypotheticals. The test — could the user verify this by reading something you just produced or did? If yes, no evidence needed. If they'd have to trust your word about the outside world, store evidence.
Query previous sessions to retrieve context, decisions, code changes, or other information. Use when you need to search for context about past conversations and work.
Web search using Jina Search API. Returns search results with titles, URLs, and descriptions. Use for finding documentation, facts, current information, or any web content. Lightweight, no browser required.
| name | visit-webpage |
| title | Visit Webpage |
| description | Visit a webpage and extract its content as markdown, or fetch images. Use for reading articles, documentation, or any web page content. Handles both HTML pages (via Jina Reader) and image URLs (downloads and saves locally). |
Fetch and extract readable content from web pages as markdown, or download images. Handles JavaScript-rendered content via Jina Reader service.
Optionally set JINA_API_KEY in your shell environment for higher Jina Reader rate limits.
Without an API key, the service works with rate limits.
python3 {baseDir}/visit.py <url>
When this skill is loaded, resolve {baseDir} to this skill directory:
/Users/kieran/.pi/agent/skills/visit-webpage
# Read an article (returns markdown)
python3 {baseDir}/visit.py https://example.com/article
# Fetch documentation
python3 {baseDir}/visit.py https://docs.python.org/3/library/asyncio.html
# Download an image (auto-detected by content-type)
python3 {baseDir}/visit.py https://example.com/image.png
# Then use read tool to view: read /tmp/visit-image-xxx.png
For HTML pages: Returns markdown content to stdout.
For images: Downloads the image to a temp file and prints the path. Use the read tool to view it. Supports PNG, JPEG, GIF, and WebP formats.
tee/sed by default; Pi already truncates long command output and stores full output when needed./tmp/*.md file after confirming the command works.read to view)