agentleFS
gdhub-iogdhub-io/character-forge-mcp/skills/translate-character/SKILL.md

translate-character

Skill1 starsChanged 4 days agoMITView on GitHub

Translate a saved character-forge character into another language (Russian for now) — the whole card, written as the character's editor would write it in that language, and the voice monologue spoken in the same voice. English stays the original. Use when the user asks to translate a character, add a language, or make a Russian version.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Keep exactly: the structure and every list's length, numbers, #colours, lowercase ids and enums (film, left_shoulder, knock…), names of …

    line 13
  2. Very specific:

    Only if the character already has a voice (assets/voice/voice.json).

    line 18
  3. Very specific:

    Add the translated kicker and blurb under i18n.<lang> in site/src/showcase.json (blurb ≤ 160 characters, written to fit, never cut), then …

    line 25

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Translate a character

English is always the source. A translation is an extra layer next to it: card.<lang>.json and assets/voice/reference.<lang>.mp3. Requires the character-forge MCP server.

1. Card — free

  1. character_get the character and read the whole card.
  2. Translate every free-text field into the target language. Write it as the editor would in that language: plain, clear narration a stranger understands on first read, in a voice tuned to the character, not a word-for-word copy. The character's own lines (pet words, sample lines, inner monologue, what others say) keep their manner: slang stays slang, a stammer stays a stammer, broken grammar stays broken.
  3. Keep exactly: the structure and every list's length, numbers, #colours, lowercase ids and enums (film, left_shoulder, knock…), names of people and places unless the language has a settled form.
  4. character_translate with id, lang and the translated card. It rejects any change to the structure or fixed values; fix and resend.
    • The card is long: in Claude Code, hand the translation to a subagent with the card path and these rules, have it write the JSON to a file, then save it with character_translate (or the CLI: character-forge translate <id> <file> --lang ru).

2. Voice — paid, ask first

Only if the character already has a voice (assets/voice/voice.json).

  1. Read prompts/voice.md ("Sample script") and prompts/scene.txt / prompts/idle.txt. Write the monologue natively in the target language for the moment the video shows (same place, pose, beat and mood), in the character's own way of speaking. Not a word-for-word copy: it must sound like a person talking, and a stranger must understand it.
  2. Tell the user the price (about the same as the original line: ElevenLabs ≈ $0.10 per 1000 characters, Qwen for minors ≈ $0.03) and wait for a yes.
  3. Translate the voice description too, by meaning, not word for word (who they are, how they talk, age, timbre, mood). character_voice_translation with id, lang, the monologue and the description generates a new voice in that language from that prompt → assets/voice/reference.<lang>.mp3.
    • New voices: character_voice takes translations: [{ lang, text }] and records every language in one call. Each language is performed fresh from the voice description, not copied from the English take.

3. Website — if the character is on the site

Add the translated kicker and blurb under i18n.<lang> in site/src/showcase.json (blurb ≤ 160 characters, written to fit, never cut), then npm run sync in site/.

4. Report (in the user's language)

What was translated, the voice file, the cost, and anything that couldn't be translated faithfully.

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.