use-ai-i18n-mcp
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests translation content and metadata from the local project via MCP tools such as
ai_i18n_list_translations(documented inSKILL.mdandreferences/tool-contracts.md). Since this data originates from the source code and translation files, it could contain instructions designed to influence the agent's behavior. The agent possesses capabilities to modify files usingai_i18n_set_translationsand to execute project-defined scripts via Vite builds. While the skill mandates user approval for deletions and overwriting existing values, the batch processing of translation data remains an ingestion point for untrusted content. - [COMMAND_EXECUTION]: The agent is instructed to "Run the selected app's full Vite Build" to generate or update extraction files (referenced in
SKILL.mdandreferences/recovery.md). This involves the execution of build scripts defined within the local project environment. - [EXTERNAL_DOWNLOADS]: The skill directs the agent to fetch documentation from
https://bosens-china.github.io/ai-i18n/llms.txtfor guidance on registration and troubleshooting. This resource is hosted on the skill author's GitHub Pages domain and is used to retrieve static instructional content.
Audit Metadata