Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/tale-project/tale/write-translationsnpx skills add tale-project/tale --skill write-translationsgit clone --depth 1 https://github.com/tale-project/taleWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/tale-project/tale/write-translations)<a href="https://agentmods.dev/skills/tale-project/tale/write-translations"><img src="https://agentmods.dev/badge/skills/tale-project/tale/write-translations.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00140 | $0.02424 |
| Opus 5 | $0.00070 | $0.01212 |
| Sonnet 5 | $0.00028 | $0.00485 |
| Haiku 4.5 | $0.00014 | $0.00242 |
Grade A, and why
write-translations scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 144 lines — stays where its author put it; the contents beside it link to each section on GitHub.
write-translations
Tale ships one calm, opinionated, second-person-informal narrator in three languages — the German page
is the same voice written natively in German, never a German rendering of the English. This file is the
cross-locale contract; the data that changes between languages (strike lists, drift patterns, gender
maps, formal-pronoun denylists) lives in the framework's per-locale test data under
packages/ui/src/i18n/tests/locales/ and the
per-locale voice files. Read this first, then the locale file for the locale you're in.
When this applies
Editing any non-English value under services/*/messages/ (e.g. de.yml, fr.yml),
packages/ui/src/i18n/messages/, or any page under docs/<locale>/. Then read
locales/<locale>/AGENTS.md for that language's voice doctrine and drift catalogue. The
two reliable failure modes are bureaucratic German (passive present, sentence-final erfolgreich,
third-person Sie) and marketed French (Découvrez, N'hésitez pas à, stacked nominal phrases) —
both translate the words and lose the voice.
Write a note first
Invoke write-notes and record your answers to this form before you translate:
- Locale & files: Describe the locale and files, and the source meaning you must convey (not the words).
- Voice risk: Describe this locale's drift mode to avoid (bureaucratic German / marketed French) and how you'll keep the native voice.
- Must-match: Describe the UI labels and compound terms that must stay exact, and how you confirmed them against the shipped strings.
- Risks & unknowns: Describe where the translation might read non-natively or drift from the source meaning.
The rules
These four fail review. The first is reviewer-caught (voice doesn't lint cleanly); the rest are enforced by the i18n test suite (see Patterns).
- Same voice across locales. Translation is a rewrite of the same narrator in another language, not a render of the source words — translate meaning, not words. Sentence structure, idiom, and noun choice all differ: the German equivalent of an English three-clause sentence is often one sentence with a verb-final subordinate clause; the French equivalent of a stacked English noun phrase is often a relative clause. A page that reads calmly in English and bureaucratically in German has a tone bug; fix the wording. The drift modes are language-specific — your per-locale file names yours. (reviewer-caught)
What ships with it
16 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- BUCKETS.md 6.2 KB
- CONVENTIONS.md 7.0 KB
- GLOSSARY_GUIDE.md 3.7 KB
- locales/de-CH/AGENTS.md 3.1 KB
- locales/de-CH/examples.md 937 B
- locales/de/AGENTS.md 6.9 KB
- locales/de/drift-catalogue.md 4.5 KB
- locales/de/examples.md 4.1 KB
- locales/de/glossary-notes.md 774 B
- locales/en/AGENTS.md 4.6 KB
- locales/en/examples.md 2.9 KB
- locales/en/glossary-notes.md 209 B
- locales/fr/AGENTS.md 6.1 KB
- locales/fr/drift-catalogue.md 3.2 KB
- locales/fr/examples.md 3.1 KB
- locales/fr/glossary-notes.md 953 B
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 144 lines · 140 tokens per session scan A 9f78c4f0da64
write-translations is a skill published in the GitHub repository tale-project/tale (25 stars, last pushed today), licensed MIT. It adds 140 tokens to every session and 2,424 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
dsh-translate-docs
Manually run the extended DeepSeek Harness bilingual-document workflow, including generated briefings, delegated prose translation, whole-document translation, and scoped pairing verification.
gpu-document-processing
Use when processing large PDFs, document collections, or bulk text extraction tasks that benefit from GPU-accelerated processing. Triggers when the user provides large documents or needs bulk document analysis.
potpie-source-ingestion
Use when the user explicitly asks to ingest, refresh, or deeply understand a repository, PR, issue, ticket, runbook, incident report, document, or web link into Potpie. The harness performs todo-driven discovery, uses local/GitHub/integration tools and read-only subagents when available, builds evidence-backed…
om-integration-builder
Build integration provider packages for the Open Mercato Integration Marketplace (payment, shipping, data-sync, webhook). Scaffolds the npm package, adapter, credentials, widget injection, webhook processing, health checks, i18n, tests. Triggers on "build integration", "add provider", "integrate with…
enrich-excel
Enrich a live Excel spreadsheet by researching missing data with TinyFish and writing results directly into the running Excel app via AppleScript. Use when a user says "fill in my spreadsheet", "enrich this Excel file", "research and fill missing data", "update my Excel sheet", "look up data for each row", or any…
translate-doc-zh
Translate an English document under docs/en/ into the matching Chinese document under docs/zh/.