Skip to content
Back to search

claude-skills

SkillWorks with Claude CodeGitHub

My Claude Code skills, with the scripts that run them unattended. They are written for my machine (Arch Linux, paths under /home/kim) and my habits, so expect to adapt them rather than use them as…

Safety scan

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

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/kim-nejudne--claude-skills/badge.svg)](/listings/kim-nejudne--claude-skills)

HTML

<a href="/listings/kim-nejudne--claude-skills"><img src="/api/listings/kim-nejudne--claude-skills/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.

git clone git@github.com:kim-nejudne/claude-skills.git ~/Documents/claude-skills
~/Documents/claude-skills/install.sh           # links skills/* to ~/.claude/skills, bin/* to ~/.local/bin
~/Documents/claude-skills/install.sh --check   # report missing or wrong links

A real file or folder already at a target is moved aside to `<target>.pre-install`, never deleted.

**Tools used:** `uv` (each script fetches its own Python packages), `python3`, `poppler` (`pdftotext`, `pdftoppm`), `java` (for epubcheck, downloaded on first use), `chromium` (screenshots), `jq`, `tmux` and `notify-send`.

**Environment** (bootstrap-autopilot, droplet deploys only), e.g. in a shell file that isn't committed:

export BOOTSTRAP_DROPLET_IP=…   # the server apps deploy to
export BOOTSTRAP_DOMAIN=…       # apps live at <app>.$BOOTSTRAP_DOMAIN

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 · 82 lines
  1. # claude-skills
  2. My [Claude Code](https://claude.com/claude-code) skills, with the scripts that run them unattended.
  3. They are written for my machine (Arch Linux, paths under `/home/kim`) and my habits, so expect to
  4. adapt them rather than use them as they are.
  5. ## Skills
  6. Every skill except `karpathy-guidelines` runs only when you type it (`/name …`); none triggers on
  7. its own.
  8. **Nonfiction books (PDF)**
  9. - `split-book`: splits a book PDF into one PDF per chapter, with each chapter's endnotes, and
  10. writes a manifest the other two reuse.
  11. - `reading-companion`: writes a study companion PDF per chapter (context, argument map, close
  12. reading, glossary, who's who, quotations), with a separate subagent fact-checking each one.
  13. - `companion-epub`: builds one phone-friendly EPUB that puts each companion before its chapter's
  14. reflowed text.
  15. **Fiction (EPUB)**
  16. - `fiction-split`: finds a novel's or story collection's chapters in the EPUB and drops front and
  17. back matter.
  18. - `fiction-companion`: writes a spoiler-safe, text-only companion for each chapter: Before you
  19. read, then After you read. The companion for chapter N uses only chapters 1..N, enforced by
  20. per-chapter digests and a separate spoiler checker. The book gets a spoiler-free intro and a
  21. full-spoiler end essay.
  22. - `fiction-epub`: builds one EPUB holding, for each chapter, Before you read, the source EPUB's own
  23. chapter XHTML, then After you read. It is checked with a word diff against the source,
  24. epubcheck and screenshots.
  25. **Apps**
  26. - `bootstrap-autopilot`: turns an app spec into a repo with a roadmap, deploy kit and an
  27. unattended build loop. Deploys go to a droplet or Vercel, or there is no deploy at all.
  28. **General**
  29. - `karpathy-guidelines`: coding guidelines against common LLM mistakes. It is vendored from
  30. [multica-ai/andrej-karpathy-skills](https://github.com/multica-ai/andrej-karpathy-skills), MIT
  31. (see its `LICENSE`).
  32. ## Autopilots (`bin/`)
  33. `reading-autopilot` and `fiction-autopilot` run a whole book unattended:
  34. ```
  35. fiction-autopilot <book.epub | book folder> # starts its own tmux session
  36. fiction-autopilot attach
  37. ```
  38. - **Stages:** split, then companions, then EPUB, one headless `claude -p` session per stage.
  39. - **Safety:** each run has a 4 GB memory cap and keeps the machine awake.
  40. - **Usage limits:** it waits for the limit to reset, and pauses at 90% of the 7-day limit.
  41. - **Retries:** each stage gets 3 tries; any permission denial stops the run.
  42. - **Resumable:** run it again and it picks up where it stopped.
  43. ## Install
  44. ```
  45. git clone git@github.com:kim-nejudne/claude-skills.git ~/Documents/claude-skills
  46. ~/Documents/claude-skills/install.sh # links skills/* to ~/.claude/skills, bin/* to ~/.local/bin
  47. ~/Documents/claude-skills/install.sh --check # report missing or wrong links
  48. ```
  49. A real file or folder already at a target is moved aside to `<target>.pre-install`, never deleted.
  50. **Tools used:** `uv` (each script fetches its own Python packages), `python3`, `poppler`
  51. (`pdftotext`, `pdftoppm`), `java` (for epubcheck, downloaded on first use), `chromium`
  52. (screenshots), `jq`, `tmux` and `notify-send`.
  53. **Environment** (bootstrap-autopilot, droplet deploys only), e.g. in a shell file that isn't
  54. committed:
  55. ```
  56. export BOOTSTRAP_DROPLET_IP=… # the server apps deploy to
  57. export BOOTSTRAP_DOMAIN=… # apps live at <app>.$BOOTSTRAP_DOMAIN
  58. ```
  59. ## Tests
  60. ```
  61. tests/fiction-autopilot.sh # fake claude, ~30 s
  62. skills/bootstrap-autopilot/tests/run.sh # fake claude and gh, ~2 min
  63. uv run --quiet --with lxml python -I skills/fiction-split/scripts/epubsrc.py --selftest
  64. python3 -I skills/fiction-companion/scripts/check.py --selftest
  65. uv run --quiet --with lxml --with markdown python -I skills/fiction-epub/scripts/epub.py --selftest
  66. python3 -I skills/reading-companion/scripts/check.py --selftest
  67. ```

Ratings and comments

Sign in to rate this skill.

Sign in

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