sciwrite-prose
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill directs the agent to execute local Python scripts, such as
extract_text.pyandprose_audit.py, to process user documents. These scripts are intended for text analysis and document management within the local environment. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied research materials (e.g., .tex, .docx, .md, .txt) for editing and review, which introduces a potential attack surface. 1. Ingestion points: The skill ingests content from external documents via the
extract_text.pyscript. 2. Boundary markers: The instructions lack explicit delimiters or safety prompts to isolate processed document content from the agent's internal instructions. 3. Capability inventory: The skill has the capability to execute local Python scripts using the shell. 4. Sanitization: No sanitization of input text is defined before the content is read by the agent. This represents a standard surface for indirect prompt injection common to document-editing tools, but it is aligned with the skill's primary functionality.
Audit Metadata