explain-me
Turn a topic or your own materials into a single self-contained, interactive HTML report — top tabs, clickable reference drawers, dense tables, comparison heatmaps, inline SVG, in-page search, a light/dark theme toggle, and multilingual support. Use when: user says "/explain-me", "/reportme", "make a report", "create report", "explain this topic". Inspired by beret21/reportme.
- License 许可
- AGPL-3.0
- 最近更新
- 2026-11-14
- Stars
- ★ 25
- Forks
- 9
01 前言 Front Matter
- name
- explain-me
- owner
- pm
- scope
- common
- status
- experimental
- version
- 1.0.1
- metadata
- { "type": "process", "triggers": [ "/explain-me", "/reportme", "make a report", "create report", "explain this topic" ] }
- attribution
- { "note": "Inspired by beret21/reportme. All code is self-authored.", "source": "https://github.com/beret21/reportme", "license": "MIT" }
- description
- Turn a topic or your own materials into a single self-contained, interactive HTML report — top tabs, clickable reference drawers, dense tables, comparison heatmaps, inline SVG, in-page search, a light/dark theme toggle, and multilingual support. Use when: user says "/explain-me", "/reportme", "make a report", "create report", "explain this topic". Inspired by beret21/reportme.
- l2_propagate
- true
- last_reviewed
- 1785715200000
- prerequisites
- Python 3 (for validate_report.py), git + gh (for optional publish)
02 资源文件 7
- 参考文档references/BUILD_GUIDE.md33.7 KB
- 参考文档references/loanword-refinements.json3.0 KB
- 参考文档references/PLATFORM_HARNESS.md23.8 KB
- 脚本scripts/validate_report.py16.1 KB
- 其他templates/publish/.nojekyll—
- 其他templates/publish/dot-gitignore176 B
- 参考文档templates/report.html30.2 KB
03 SKILL.md 正文
Skill: explain-me
Context
Generates a single self-contained interactive HTML report from a topic (research mode) or existing materials (render mode). The entire output is one HTML file with all CSS, JS, and data inlined — it opens directly from disk (file://) with no server or build process required.
Inspired by beret21/reportme v0.4.1 (MIT license). All code is self-authored. Supports all 4 workspace platforms: Claude CLI, Claude Desktop App, Antigravity (VS Code), and Antigravity CLI (Gemini CLI).
Subcommands
| Command | Description |
|---|---|
new (default) |
Full end-to-end report creation from a topic |
render |
Turn existing local or remote data into report format (no research) |
add |
Append components (tabs, subtabs, sections, heatmaps, figures, references) to an existing report |
update |
Conversationally revise and re-verify an existing report |
open |
Launch the report in a browser for manual inspection |
verify |
Run structural consistency and content/language reviews |
publish |
Optional push to GitHub Pages (requires git + gh) |
help |
Show usage information |
Usage
/explain-me [subcommand] [options] "<topic or materials>"
Options
| Option | Description | Example |
|---|---|---|
--lang |
Output language (defaults to user's prompting language) | --lang en, --lang ko |
--langs |
Multilingual output (space-separated, max 5, recommended ≤3) | --langs ko en ja |
--depth |
Report depth: brief / standard (default) / deep |
--depth deep |
--source |
Source mode: research (default) or data (auto-detected) |
--source data |
--tabs |
Custom tab list (comma-separated) | --tabs "Overview,Analysis,References" |
--out |
Output file path | --out ./reports/market-2026.html |
--no-verify |
Skip verification (FORBIDDEN in this workspace) | --no-verify |
--no-svg |
Skip inline SVG diagrams | --no-svg |
--repo |
Publish: repository slug | --repo ev-report-2026 |
--owner |
Publish: GitHub account | --owner beret21 |
--private |
Publish: private repository (requires GitHub Pro) | --private |
--public |
Publish: public repository | --public |
Safety Model
This skill only creates and edits — it never deletes your files or content.
- Network: Only used for read-only research (WebSearch/WebFetch). No outbound connections from the output HTML.
- Git/gh: Only invoked for the explicit
publishsubcommand, after a secret scan passes. - Temporary files: Confined to OS temp directory.
- Secrets: §11 of BUILD_GUIDE scans for API keys, tokens, passwords before delivery.
Workflow Summary
- Parameter confirmation — confirm TOPIC, LANG, DEPTH, TABS with user
- Content acquisition — research (WebSearch/WebFetch) or data (Read user files)
- Structure proposal — propose tab layout, get user approval
- Template copy + fill — copy
templates/report.html, replace placeholders - Content writing — fill tabs with verified facts using design system components
- Multi-agent verification — §6 content (4 personas + 4 lenses), §7 language proofreading, §7.5 SVG diagrams
- Structural validation —
scripts/validate_report.py(must return 0 errors) - Browser verification — test tabs, drawer, search, theme, mobile,
file://self-sufficiency - Delivery — output single HTML file; optional
publishto GitHub Pages
Bundled Assets
| Asset | Location | Purpose |
|---|---|---|
| BUILD_GUIDE.md | references/BUILD_GUIDE.md |
Complete report building engine (16 sections) |
| PLATFORM_HARNESS.md | references/PLATFORM_HARNESS.md |
Platform-specific subagent dispatch patterns |
| loanword-refinements.json | references/loanword-refinements.json |
Korean loanword refinement data for §7 proofreading |
| validate_report.py | scripts/validate_report.py |
HTML structural validator (stdlib Python) |
| report.html | templates/report.html |
Single-file interactive HTML template |
| .nojekyll | templates/publish/.nojekyll |
GitHub Pages Jekyll bypass |
| dot-gitignore | templates/publish/dot-gitignore |
Gitignore template for published repos |
Platform Support
All 4 workspace platforms are supported. Tool names differ per platform — see references/PLATFORM_HARNESS.md for platform-specific subagent dispatch patterns.
| Platform | Shell | Subagent | File I/O |
|---|---|---|---|
| Claude CLI | Bash |
Agent() |
Read/Write/Edit |
| Claude Desktop App | Bash |
Agent() (limited Teams) |
Read/Write/Edit |
| Antigravity (VS Code) | run_command |
invoke_subagent |
view_file/write_to_file |
| Antigravity CLI | run_command |
invoke_subagent |
view_file/write_to_file |
Kickoff
For new report generation, read references/BUILD_GUIDE.md in full first, then follow it exactly. Use references/PLATFORM_HARNESS.md for platform-specific subagent dispatch patterns.
04 同仓库更多 Skills
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.
i18n-layout
Text layout and encoding guidance: character encoding (UTF-8, legacy Korean code pages, BOM hazards), line endings, RTL/bidi handling, and script-specific font selection (Hangul focus). Use when: encoding is corrupted or mojibake appears, an RTL locale must render correctly, fonts must be chosen for a script, or CRLF/BOM issues surface.