repo-bilingual-readme
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted text and metadata from external GitHub repositories and research paper pages during its translation and link-enrichment workflows. This ingestion of external data creates a surface where malicious instructions embedded in a repository's documentation could attempt to influence the agent's behavior.
- Ingestion points: Markdown files (
SKILL.mdrefers to existing repo files) and web pages for paper/project metadata verification. - Boundary markers: The skill contains instructions to focus on translation scope but lacks specific delimiters or 'ignore' warnings for the data being processed.
- Capability inventory: The agent can write to the repository (
git push), perform network searches, and read local project files. - Sanitization: No explicit sanitization or filtering of external content is specified before processing.
- [COMMAND_EXECUTION]: The skill uses shell-based git tools (
git status,git push,git diff) to perform its primary function of document maintenance. While these are necessary for the skill's purpose, they enable modification of remote repositories. The instructions mitigate potential abuse by prohibiting force-pushes and requiring the agent to verify the results of remote writes.
Audit Metadata