agentleFS
kaderlabskaderlabs/polyglot/SKILL.md

polyglot

Skill0 starsChanged 6 months agoMITView on GitHub

Audit and improve Claude Code skill activation.

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. simota/agent-skills★ 96same name
    2. Amal-David/polyglot★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 4 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If no language is given, ask: "Which language?

    line 19
  2. Very specific:

    If --skill <name> is passed, filter to that skill only.

    line 30
  3. Very specific:

    Write target: always when_to_use. If the skill has no when_to_use field, create one.

    line 57
  4. Very specific:

    If creating when_to_use from scratch (skill had no activation field), extract any existing "Use when" clause from description as the base, …

    line 64
  5. Distinctive:

    Use the Edit tool. One skill at a time.

    line 66
  6. Distinctive:

    Only list skills that were actually updated:

    line 70
  7. Very specific:

    After the language pass, scan every SKILL.md (not just the ones updated) and check whether any of these activation signals exist:

    line 83
  8. Distinctive:

    If any skills are missing both, append a notice to the report:

    line 90
  9. Very specific:

    If the user answers y, draft a Use when: clause for each affected skill based on its name and existing description body.

    line 101
  10. Very specific:

    Do not add "Use when" clauses to path-based skills (those already have paths frontmatter for conditional activation).

    line 103

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md
<objective> Scan your Claude Code skill library and append native-language trigger phrases to any skill whose description only covers English. Skills activate naturally when you think and write in your preferred language — without rewriting the skills themselves.

Language is required. No default. </objective>

<invocation> - `/polyglot sv` — add Swedish triggers to all skills missing them - `/polyglot fr` — French - `/polyglot sv --dry-run` — preview without writing - `/polyglot sv --skill my-skill` — single skill only

If no language is given, ask: "Which language? (e.g. sv, fr, de, es)" </invocation>

<process>

1 — Locate skills

find ~/.claude/skills -name "SKILL.md" 2>/dev/null

If --skill <name> is passed, filter to that skill only.

2 — Detect coverage

For each SKILL.md, read the when_to_use frontmatter field first. Fall back to description if when_to_use is absent.

Language markers:

  • sv — å, ä, ö
  • fr — é, è, à, ç, ù
  • de — ü, ß
  • es — ñ, ¿, ¡
  • other — look for Also ([lang]): marker

Skip if target-language characters already present in the activation field.

Skip path-based skills: activation field consists primarily of glob patterns (**/, src/, file extensions). These trigger on file edits, not words — language triggers add noise.

3 — Generate triggers

Count the existing trigger phrases. Generate the same number in the target language — no more, no fewer. Match density, don't bloat.

Trigger phrases should:

  • Match how a native speaker naturally asks for this skill
  • Cover both formal and casual registers
  • Stay true to the skill's actual domain
  • Include action verbs and question fragments where natural

Write target: always when_to_use. If the skill has no when_to_use field, create one. If triggers currently live only in description, leave description unchanged and add when_to_use instead.

Append to when_to_use using this pattern:

[existing when_to_use content]. Also: "phrase1", "phrase2", ...

If creating when_to_use from scratch (skill had no activation field), extract any existing "Use when" clause from description as the base, then append the new-language triggers.

Use the Edit tool. One skill at a time.

4 — Report

Only list skills that were actually updated:

polyglot — sv · 2026-05-26
──────────────────────────────────────
✓  my-skill      +4 triggers
✓  other-skill   +6 triggers
──────────────────────────────────────
2 updated

5 — Check for missing "Use when"

After the language pass, scan every SKILL.md (not just the ones updated) and check whether any of these activation signals exist:

  • description contains "use when", "use this skill when", or "trigger" (case-insensitive)
  • a when_to_use frontmatter field is present

Skills that pass none of these checks lack reliable activation guidance.

If any skills are missing both, append a notice to the report:

──────────────────────────────────────
⚠  Skills with no "Use when" clause:
   lint, test, frontend-design
   These won't auto-activate reliably.
   Add "Use when" descriptions? (y/n)
──────────────────────────────────────

If the user answers y, draft a Use when: clause for each affected skill based on its name and existing description body. Present each as a one-line proposal and ask for confirmation before writing.

Do not add "Use when" clauses to path-based skills (those already have paths frontmatter for conditional activation).

</process>

<dry_run> With --dry-run: print proposed additions without writing. Confirm before proceeding.

[my-skill]
  would add: "phrase1", "phrase2", ...

Proceed? (y/n)

</dry_run>

<idempotency> Running polyglot twice is safe. Coverage detection on characters prevents double-adding. </idempotency>

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.

Reviews can't be read right now.

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.