Menü
Use when creating, editing, validating, or enforcing glossary-driven terminology.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Basierend auf der SOC-Berufsklassifikation
Use when translating game content that requires cost-optimized model routing, linear workflow efficiency, and immediate formatting to avoid multi-loop review cycles
Use when high-quality translation is needed with multi-agent review and quality gates before overwriting source files.
Use when a markdown draft or docs page must be checked for structural validity and documentation style compliance before writeback or publishing.
Use when high-quality translation is needed with multi-agent review and quality gates before overwriting source files.
Use when extracted rulebook markdown needs to be split into semantic documentation files and navigation. Trigger this skill from `init-doc`, future append/add-document flows, or whenever regenerated `_pages.md` source invalidates the existing chapter map. Do not use this skill for temporary translation chunking; that belongs to a separate draft-only translation workflow.
Use when performing final quality checks before publishing the documentation site. Use when all translation is complete and you need to verify titles/descriptions are translated, check for misplaced or erroneous content, and run page-reference link checks.
| name | terminology-management |
| description | Use when creating, editing, validating, or enforcing glossary-driven terminology. |
Maintain glossary quality and enforce terminology consistency across all translated content.
Core principle: Glossary is the source of truth; validation must run before and after any terminology update.
glossary.json.uv run python scripts/validate_glossary.py
Before running candidate generation, inspect the extracted source markdown for high-signal terminology sources:
Use those sections to complete a first-pass glossary bootstrap:
Run:
uv run python scripts/term_generate.py --min-frequency 2
Then immediately pre-calculate evidence for all candidates in one pass:
uv run python scripts/term_cal_batch.py
This scans the corpus once and caches counts + context for every candidate.
Subsequent term_edit.py calls hit the cache; no re-scanning needed.
Classify each candidate as:
If proper_nouns.mode != keep_original, repeated proper nouns (>=2) must be managed.
For unmanaged terms, evidence is read from cache (populated by term_cal_batch.py):
uv run python scripts/term_edit.py --term "<TERM>" --set-zh "<ZH_TERM>" --status approved --mark-term
To inspect evidence for a single term without editing:
uv run python scripts/term_edit.py --term "<TERM>" --cal
If the corpus changed since the last batch run, re-run term_cal_batch.py to refresh the cache.
Run:
uv run python scripts/term_read.py --fail-on-forbidden
Check for:
term_read.py until clean.term_read.py is clean for that change.Stop when:
Return to Step 2 when:
Never:
Continue with /check-consistency, /translate, or /super-translate.