Translatepro
ReviewAudited by ClawScan on May 1, 2026.
Overview
TranslatePro appears to be a coherent translation/localization skill, but users should notice that it uses third-party translation APIs, API keys, local files, and translation memory.
This skill is reasonable for translation and localization tasks. Before installing or using it, confirm that Google/DeepL API use is acceptable for your content, restrict the API keys, avoid sending secrets or regulated data unless approved, and keep bulk folder paths and translation memories scoped to the intended project.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Misused or overexposed API keys could incur charges, consume quota, or allow translation requests under the user's account.
The skill requires API keys for Google Translate and DeepL, which is expected for its multi-engine translation purpose but gives the agent access to paid provider credentials.
"requires":{"env":["GOOGLE_TRANSLATE_API_KEY","DEEPL_API_KEY"],"bins":["curl","jq"]}Use restricted API keys with quotas, keep them in environment variables or a secret manager, and rotate them if exposed.
Private, regulated, or unpublished content provided for translation may be sent to Google or DeepL for processing.
The skill discloses that content may be routed to external translation providers. This is purpose-aligned, but it creates a third-party data boundary for whatever text or files the user asks to translate.
"Google Translate API": 100+ languages ... "DeepL API": Superior neural translation ... "Intelligent fallback": Routes high-stakes content to DeepL, bulk content to Google
Check provider privacy/retention terms before translating sensitive material, redact secrets, and use approved enterprise settings for regulated content.
A bad or poisoned translation memory could affect later translations, especially if fuzzy matches are accepted without review.
The skill describes persistent translation memory and optional automatic reuse of fuzzy matches, which can preserve consistency but can also propagate stale, incorrect, or untrusted translations.
"Build incremental TM": Every translation adds to your organizational memory ... "Auto-accept fuzzy matches: yes"
Use trusted TMX/XLIFF files, review fuzzy matches for important content, and separate translation memories by customer, project, or sensitivity level.
The agent could process more files than intended or overwrite/produce large numbers of localized files if paths are too broad.
The bulk localization example involves reading a folder of files and writing translated outputs. This is expected for localization, but it can touch many local documents if the user provides a broad path.
Localize my entire product website ... Input folder: /content/pages/ ... File types: .md, .html ... Output format: /localized/{language_code}/Use narrow input/output directories, keep backups or version control, and review diffs before publishing localized content.