ssci-polish
Pass
Audited by Gen Agent Trust Hub on Aug 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides structured Python code snippets that the agent is expected to execute locally to perform document manipulation. This includes reading source files, performing regex-based sentence analysis, and writing formatted changes to a new file using the
python-docxlibrary. - [INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection by processing external, untrusted content from user-provided DOCX files.
- Ingestion points: Full document text, including body paragraphs and table cells, is extracted using
python-docxand printed to the agent's context (SKILL.md, Step 1). - Boundary markers: The instructions do not define clear delimiters or warning markers to separate the document content from the agent's core instructions.
- Capability inventory: The agent is granted the ability to execute Python code and perform file system operations (read/write).
- Sanitization: There is no logic provided to sanitize or filter the content of the document before it is processed by the agent.
Audit Metadata