使用方法
- 复制安装命令,在自己的智能体环境安装前先检查源码包。
- 从原始文档核对所需工具、凭据和支持的宿主。
- 说明输入材料与期望交付物,并对照来源检查输出。
npx skills add https://github.com/terrylica/cc-skills --skill render-incident-report-image开始前需了解
- 安装不包含模型订阅、API 额度或访问文件的权限。
- 仓库星数和安装量反映活跃情况,不等于独立质量或安全测试。
按源仓库内容呈现,保留标题、案例、代码、表格、链接以及原文引用的演示图片。
render-incident-report-image
Self-Evolving Skill: This skill improves through use. If instructions are wrong, parameters drifted, or a workaround was needed — fix this file immediately, don't defer. Only update for real, reproducible issues.
Render a clean monospace report PNG via the TS core pushover_core.ts render (Satori → vector SVG → @resvg/resvg-js → PNG).
Device-calibrated facts (verified on a physical iPhone 13 mini, 2026-05-30)
- 72-column hard wrap is comfortably legible inline without pinch-zoom (40/56/72 all tested; 72 confirmed fine).
- Pushover fits images to width — it never crops horizontally. Earlier "right-edge trimmed" images were a renderer bug (no wrapping), not Pushover.
- No practical pixel-height ceiling: a 30,000 px-tall image scrolls/zooms fine, far under the 5 MB attachment cap → render the whole report in one tall image, never paginate.
- Auto-fit width (2026-06-01): 72 cols is the MAX wrap, not a fixed canvas width. The canvas is sized to the longest actual line (
maxChars × 18px advance + 48px pad, monospace ⇒ exact/deterministic), so short reports don't waste right-margin whitespace — Pushover scales the snug image up, rendering text larger. (e.g. a 43-char digest: 1344 → 822 px, ~1.6× bigger on screen.) No image post-processing needed.
Usage
env -u HTTPS_PROXY -u HTTP_PROXY bun "$(cc-plugin-root pushover-commander)/skills/_lib/pushover_core.ts" render \
--in report.txt --out /tmp/report.png # or: cat report.txt | bun .../pushover_core.ts render --out /tmp/report.png
# then attach with send-notification --attach /tmp/report.pngLine markup (first 2 chars)
| Prefix | Style |
|---|---|
# | heading (green) |
! | subheading (amber) |
> | accent (cyan) |
. | dim/muted |
| (none) | body (light grey) |
Wrapping is word-aware (greedy, fold -s semantics): lines break at spaces and NEVER mid-word. Only a single token that ALONE exceeds the column width (paths, hashes, URLs) is hard-broken so nothing overflows the canvas. Lines already within the width keep their exact spacing (aligned tables stay aligned). Greedy is the correct algorithm for left-aligned monospace — Knuth-Plass optimizes justified paragraphs and adds nothing to ragged-right fixed-width text.
Post-Execution Reflection
After this skill completes, check before closing:
- Did the PNG render with correct word-wrap and no horizontal cut-off at the target device width? If text overflowed, fix the wrap/width.
- Did the Satori to resvg pipeline complete without error? A blank or partial image means a rendering regression to fix.
Only update if the issue is real and reproducible — not speculative.

