glossary-terms-extraction
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill acts as an ingestion point for untrusted external data which is then used to perform actions.\n
- Ingestion points: The skill reads project source strings, fetches content from user-provided URLs (product docs, help centers), and processes uploaded files such as PDFs or documents.\n
- Boundary markers: The instructions do not specify technical prompt delimiters but implement a critical 'Review Gate' where every proposed term, definition, and translation must be manually approved or edited by the user.\n
- Capability inventory: The skill possesses the ability to create and update glossaries, concepts, and terms through Crowdin MCP tools and the Crowdin CLI.\n
- Sanitization: Manual review is the primary control mechanism to ensure that no malicious instructions or incorrect data are written to the glossary.\n- [CREDENTIALS_UNSAFE]: The documentation references authentication mechanisms that involve sensitive credentials.\n
- Evidence: The file
references/cli.mdmentionsCROWDIN_PERSONAL_TOKENand.envfiles.\n - Context: The skill emphasizes secure handling by stating that token values should never be echoed or written and should be stored in gitignored
.envfiles to prevent accidental leakage, aligning with secure development practices.
Audit Metadata