Skip to content
Back to search

kernel

Visualize agent skills and documentation as architecture cards with a reading order. Parses SKILL.md into cards, groups, and citations so you know where to start.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/puti967--kernel/badge.svg)](/listings/puti967--kernel)

HTML

<a href="/listings/puti967--kernel"><img src="/api/listings/puti967--kernel/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

These steps run a script from the network that findsafeskills does not scan.

Requires Node.js 18 or newer. One command installs the skill.

macOS or Linux:

curl -fsSL https://raw.githubusercontent.com/puti967/kernel/main/install.sh | bash

Windows Command Prompt:

powershell -NoProfile -ExecutionPolicy Bypass -Command "iex (irm https://raw.githubusercontent.com/puti967/kernel/main/install.ps1)"

The installer puts the same skill on Claude Code, Codex, Cursor, and Grok. Open a new chat, then run one command. Kernel prints a `page` path, writes that HTML file, and opens the card map.

- Claude Code: `/kernel /path/to/skill` - Codex: `/kernel /path/to/skill` - Cursor: `/kernel /path/to/skill` - Grok: `/kernel /path/to/skill`

Replace `/path/to/skill` with the skill directory.

A directory is a skill only when it has `SKILL.md` with `name` and `description`. Otherwise Kernel stops and does not draw a map.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 80 lines · 2 findings marked
  1. # Kernel
  2. Visualize an agent skill, or any unfamiliar document set, as architecture cards. Kernel shows where to start, what to read next, and why the author connected one file to another.
  3. You do not need to guess the outline. A citation in the text is a connection. A file with no citation is still its own card.
  4. ![A selected route marks 1 先看 and 2 下一步, and fades the other group](docs/preview.png)
  5. Select one route. Kernel marks what to read first and what comes next. Cards and lines from the other route fade. The two dots on a card are its other routes.
  6. ## Who it is for
  7. Use Kernel when you open someone else's agent skill, `SKILL.md`, or a folder of docs and do not know the reading order.
  8. It is a local Node.js tool. It writes one HTML page. It does not call a model to split files.
  9. ## Features
  10. - Parse a skill directory into architecture cards
  11. - Build a reading order from citations, not from a guessed outline
  12. - Keep uncited files as their own cards, outside the entry reading group
  13. - Mark a fan-out as a menu when the files are parallel, not a numbered sequence
  14. - Click a color marker to show one route, and only that route's lines
  15. - Open the generated page in a browser and read the original markdown
  16. ## Quick start
  17. Requires Node.js 18 or newer. One command installs the skill.
  18. macOS or Linux:
  19. ```bash
  20. curl -fsSL https://raw.githubusercontent.com/puti967/kernel/main/install.sh | bash
    • Note: Pipes a downloaded script directly into a shell (curl | sh / wget | bash)
    • Line 33 on GitHub
  21. ```
  22. Windows Command Prompt:
  23. ```cmd
  24. powershell -NoProfile -ExecutionPolicy Bypass -Command "iex (irm https://raw.githubusercontent.com/puti967/kernel/main/install.ps1)"
  25. ```
  26. The installer puts the same skill on Claude Code, Codex, Cursor, and Grok. Open a new chat, then run one command. Kernel prints a `page` path, writes that HTML file, and opens the card map.
  27. - Claude Code: `/kernel /path/to/skill`
  28. - Codex: `/kernel /path/to/skill`
  29. - Cursor: `/kernel /path/to/skill`
  30. - Grok: `/kernel /path/to/skill`
  31. Replace `/path/to/skill` with the skill directory.
  32. A directory is a skill only when it has `SKILL.md` with `name` and `description`. Otherwise Kernel stops and does not draw a map.
  33. ## How a map is built
  34. 1. Keep `.md` and `.json` files. Skip `package.json`, `test/`, and `corpus/`.
  35. 2. A citation is a markdown link or a backtick path that resolves to a file in the skill.
  36. 3. The entry reading group contains only the files `SKILL.md` cites directly.
  37. 4. Other cited files form their own groups. A file that nothing cites becomes a single card.
  38. 5. A group is a sequence only when each next card is cited by the one before it. Otherwise the cards are a menu: read the first, then any of the others.
  39. 6. One card can sit on more than one route. The color marker selects the route. Lines outside that route turn gray.
  40. ## Example
  41. Given a skill whose `SKILL.md` links to `references/sheet.md` and `references/spec.md`, Kernel draws three cards. The entry card is “read this first.” The two references are available next to it, without a fake step 2 and step 3, unless one file actually cites the other.
  42. Click the entry card to see that route. Click a color marker on a card that belongs to two groups to switch routes.
  43. ## Project layout
  44. | Path | Purpose |
  45. | --- | --- |
  46. | `bin/kernel.mjs` | Read a skill directory and write the card page |
  47. | `bin/atlas-rules.mjs` | Decide which files become cards, and which route a click opens |
  48. | `viewer/index.html` | The card page. Generated pages copy this file |
  49. Generated pages are written to `pages/`. They are output, not source.
  50. ## License
  51. [MIT](LICENSE)

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.