网页截图导出 · Web Capture Export
给一个静态 HTML 页面(或任意网页源文件)加一个右下角「导出图片」按钮,点击即用 modern-screenshot 把主体内容导出为 PNG。核心保证:截图与浏览器实际渲染一致 (文字不会因为容器宽度测量错误而换行),输出 2x 分辨率适合网络传播。
Triggers
Activate when
- 用户说“给这个网页加一个截图/导出图片按钮”。
- 用户说“网页一键导出 PNG / 长图 / 分享图”。
- 用户想把一个 HTML 报表、数据看板、落地页变成可分享的图片。
- The user asks to "add a screenshot export button" or "export this page as PNG".
Do not activate when
- 要把内容导出为 HTML/MD/DOCX/PDF 交付包;使用
lov-rich-export。 - 要把 HTML 转成 PPTX;使用
lov-html2pptx。 - 要把 PNG 转 SVG / PDF 转 PNG;使用
lov-png2svg/lov-pdf2png。 - 只是截图当前浏览器视口(无需修改页面源码);这不是本 Skill 的确定性集成任务。
User Profile (cross-session)
Read skill.yaml on every invocation. Resolve user, brand, workspace, shared
preferences, and skills.lov-integrate-modern-screenshot records without copying
personal values into this portable source. Persist only direct durable user
statements through scripts/profile_store.py, then report the saved Profile path.
See references/user-profile.md.
Skill Group Composition
Read references/skill-composition.md before composing adjacent capabilities.
Sibling Skills are optional artifact handoffs and never hidden dependencies.
Required resources
Resolve SKILL_DIR from the active Skill context, then verify:
$SKILL_DIR/scripts/integrate_screenshot.py$SKILL_DIR/assets/modern-screenshot.js$SKILL_DIR/references/integration-contract.md$SKILL_DIR/references/acceptance-checklist.md
Workflow (MANDATORY)
Step 0: Resolve skill root, dependencies, and runtime context
- Use
SKILL_DIRif provided; otherwise infer the installed skill directory. - Confirm
scripts/integrate_screenshot.pyandassets/modern-screenshot.jsexist. - Read the shared Profile via
scripts/profile_store.py read --skill-id lov-integrate-modern-screenshot --prettywhen a Profile is configured. - Do not overwrite an existing target without explicit authorization.
Step 1: Inspect the target HTML
Read the target file and identify:
- the主体内容边界 (default:
<main>plus its preceding top-level<header>; otherwise<article>; otherwise<body>contents); - any CSS design tokens (
--surface,--ink,--line,--accent) so the injected button adapts to light/dark theme; the snippet already falls back to neutral colors when tokens are absent; - the download filename (from
<title>when present).
If the intended capture region differs from the default, tell the user the exact
--selector (tag / #id / .class) to use.
Step 2: Integrate the export button
python3 "$SKILL_DIR/scripts/integrate_screenshot.py" integrate <html> \
--selector <css> --width 760 --scale 2 --title "<下载文件名>"
--widthsets the fixed capture width (default 760px;0follows content).--scalesets output resolution (default 2, suitable for sharing).- Without
--out, the source file is rewritten in place. Prefer a copy or--outwhen the user wants to keep the original.
The script embeds modern-screenshot inline (self-contained, offline), wraps the
capture region in <div id="capture">, and injects the floating button plus a
copyable error hint.
Step 3: Verify end-to-end
python3 "$SKILL_DIR/scripts/integrate_screenshot.py" verify <html> \
--scale 2 --chrome "/path/to/chrome"
The verifier renders the page in headless Chrome and asserts that the exported
PNG size equals node.scrollWidth/Height × scale exactly. A size mismatch is the
signal that the container width was measured wrong and text may reflow — fix the
--width or --selector and re-run.
Apply every item in references/acceptance-checklist.md. Report the real input,
the exact command, the OK out=WxH … result, and the output path. Do not invent
a passing verification or a score.
Step 4: Report and record
Report the output path, selector, width, scale, and verification result. Record a
real case in cases/cases.json only when a genuine integration occurred; do not
manufacture one.
Output contract
Return:
- the target HTML path and whether it was rewritten in place or to a new file;
- the capture selector, fixed width, and scale used;
- the headless-Chrome verification result (
OK out=… node=… expect=…); - any remaining gaps (e.g. web-font embedding, animated content, cross-origin media).
Dependencies
- Runtime: Python 3.8+ (stdlib only — no PyYAML needed for the CLI itself).
- Verification: headless Chrome/Chromium (
--headless=new --dump-dom). - The target page needs no build step;
modern-screenshotis embedded inline.
