agentleFS
Sophon-KrSophon-Kr/thai-native/skills/thai-polish/SKILL.md

thai-polish

Skill0 starsChanged 22 days agoMITView on GitHub

Polish Thai text so it reads natural instead of translated or AI-written. Runs the thai-native lint, fixes the flagged spans and clear translationese, keeps meaning and tech terms. Pass a file path or text; with no argument it polishes your most recent Thai output.

The file

Files
SKILL.md

thai-polish

Target: $ARGUMENTS (If empty, polish the most recent Thai text you wrote in this conversation.)

Script: python3 "${CLAUDE_SKILL_DIR}/../thai-writing/scripts/thai_native.py" (the thai-writing skill folder sits next to this one).

  1. Load the thai-writing skill first if it is not loaded in this session. It holds the rules, register presets and references.
  2. Work out genre and register from the arguments, the file type or the content. Infer; do not ask unless the reader is truly unknown and it changes the result.
  3. Lint: … lint --genre <genre> [--register <r>] --file <path>. For pasted text, pipe it on stdin with a quoted heredoc. For a file, also run … compare --snapshot <path> before editing.
  4. Read ../thai-writing/references/ai-isms.md. Fix (a) each flagged span that is a real problem and (b) translationese the lint cannot see (sentence structure, calques, wrong register). Leave sentences that already read naturally. Keep facts, numbers, names, code, links, terms per the glossary and the author's intent. Do not add content. Keep edits local: the gate in step 5 reverts an unflagged paragraph rewritten by more than about a third.
  5. Re-run lint. Aim for no error/warn; info items are judgment calls. Then gate the revision against the original, once: file → … compare --revised <path> --merged <path> (restores paragraphs marked revert); pasted text over ~800 Thai characters → save both versions to temp files, run … compare --draft <original> --revised <polished> and restore the revert paragraphs yourself. Recheck review paragraphs.
  6. Output
    • File: edit in place, then summarize the changes in 1–3 lines.
    • Text: return the polished text, then up to 5 short Thai bullets on the main changes.
    • --fix only applies safe spelling fixes; use it before step 4 if the lint shows SP errors.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.