Verify non-English phrases in your manuscript. Checks grammar, period accuracy, dialect, and translations.
View on GitHubhowells/fiction
fiction
skills/language/SKILL.md
January 25, 2026
Select agents to install to:
npx add-skill https://github.com/howells/fiction/blob/main/skills/language/SKILL.md -a claude-code --skill languageInstallation paths:
.claude/skills/language/Run the language-checker agent to verify all foreign language phrases in your manuscript. ## What This Does 1. Scans all chapters for non-English text 2. Identifies the language of each phrase 3. Verifies grammatical correctness 4. Checks period/historical appropriateness 5. Validates dialect and register (formality level) 6. Confirms translation accuracy where translations are provided 7. Resolves author markers (`**check**`, `[verify]`, etc.) ## Usage ``` /fiction:language # Check all chapters /fiction:language 5 # Check chapter 5 only /fiction:language swedish # Focus on Swedish phrases only /fiction:language 3-7 # Check chapters 3 through 7 ``` If arguments provided: $ARGUMENTS ## What It Catches - **Grammatical errors** — Spelling, agreement, conjugation, case endings - **Anachronisms** — Modern phrases in historical settings - **Dialect mismatches** — Stockholm Swedish used where Gotland dialect expected - **Register errors** — Wrong formality level (du/ni, tu/vous, du/Sie) - **Translation issues** — English translations that miss nuance - **Diacritical marks** — Missing or incorrect accents (o, u, e, etc.) ## Languages Supported The agent can verify phrases in any language by using web research to cross-reference grammar rules, dictionaries, and native speaker resources. Common languages in fiction: - Swedish, Norwegian, Danish, Finnish - German, French, Italian, Spanish, Portuguese - Russian, Polish, Ukrainian - Yiddish, Hebrew - Japanese (romanized or characters) - Any language with online resources ## Parallel Processing When checking a full manuscript, chapters are processed in parallel: 1. Spawn reader agents to extract all foreign phrases from each chapter 2. Aggregate findings 3. Verify each unique phrase (avoids duplicate research) 4. Generate comprehensive report ## Output A report with: - Summary statistics - Issues found with suggested corrections - Verified phrases (confi