Elvish Translation Tools

A Claude Code skill for translating, reviewing, and maintaining translations into Tolkien's Elvish languages (Sindarin and Quenya). Every choice is grounded in the Eldamo lexicon (Paul Strack); the skill won't propose vocabulary it can't cite.

This repo is the skill itself plus the data and tools it relies on. The skill is reusable across any translation project; project-specific workspaces (translation files, project glossaries) live in their own repos and reference this one.

Install

Clone into your Claude Code skills directory:

git clone https://github.com/joeyparrish/elvish-translation-tools \
  ~/.claude/skills/elvish-translation-tools

Claude Code discovers the skill via the SKILL.md at the repo root. The skill auto-loads when you work on Elvish translations; you can also invoke it explicitly. No further configuration is required.

To update, git pull in ~/.claude/skills/elvish-translation-tools/.

What the skill does

• Translate: takes an English string and proposes Sindarin (or Quenya) candidates, ranked by attestation, with per-element provenance and rationale.
• Review: reads an existing translation, checks every word against the lexicon, and surfaces false friends (Quenya-as-Sindarin), deprecated coinages, wrong-meaning attested forms, and missing mutations.

The skill stays honest about confidence. Anything coined is flagged as coinage; anything reconstructed is flagged as Neo-Sindarin with the original creator cited.

Worked example

For a complete example of a translation project built with this skill, see my full translation of Shaka Player's UI into Sindarin.

Standalone use of the tools

All scripts work without invoking the skill, useful for ad hoc queries or batch processing of translation files:

# Attestation lookup over the Eldamo lexicon
python3 scripts/lookup.py check lambë
python3 scripts/lookup.py search halt
python3 scripts/lookup.py forms daro

# Romanized -> CSUR Tengwar (default: \uXXXX escapes)
python3 scripts/transliterate.py sjn "daro dîn"
python3 scripts/transliterate.py qya "namárië"

# Bulk-regen tengwar fields in a schema-following YAML
python3 scripts/regen_tengwar.py path/to/translations.yaml

# Render escape-encoded Tengwar to literal PUA for a font-aware terminal
echo '' | python3 scripts/preview.py -

See scripts/README.md for the full subcommand reference and references/README.md for the grammar / mutation / schema / codepoint documentation.

Layout

• SKILL.md -- the skill itself; loaded by Claude Code.
• data/ -- compact TSV extracts of the Eldamo lexicon (committed, so the tools work out of the box) and Tecendil-format mode files for Sindarin / Quenya / Beleriand transliteration.
• scripts/ -- extract.py (regenerate TSVs from upstream Eldamo), lookup.py (attestation), transliterate.py (romanized -> CSUR), regen_tengwar.py (bulk-regen YAML tengwar fields), preview.py (escape -> literal for font-aware display), test_transliterate.py (regression suite against empirical Tecendil output).
• references/ -- hand-written grammar, mutation, CSUR codepoint, and translation source-file schema references, plus instructions for cloning upstream Eldamo when the TSVs need to be regenerated.

Requirements

• Python 3 (standard library only for lookup scripts; others use PyYAML).
• Claude Code for the skill itself; the scripts work standalone.

Related

• Eldamo (Paul Strack), CC-BY 4.0; the source the TSVs are derived from. Eldamo covers Quenya, Sindarin, Noldorin, Gnomish, and earlier conceptual stages with reliability markers.
• Tecendil (Arno Gourdol); the transliterator whose deployed rule semantics this skill's transliterate.py reproduces. Mode files vendored from arnog/tecendil-js (MIT).