ml-paper-writing
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches LaTeX style files, conference templates, and academic metadata from well-known scholarly sources and official conference domains.
- [COMMAND_EXECUTION]: Uses standard system utilities and LaTeX build tools to analyze research repositories and compile manuscripts.
- [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: Processes research codebase content and external paper abstracts retrieved from academic APIs.
- Boundary markers: Includes instructions to the agent to use explicit placeholders like
[CITATION NEEDED]and[CLAIM NEEDS EVIDENCE]for any unverified data. - Capability inventory: Capabilities include repository exploration (find, grep) and network access to scholarly databases.
- Sanitization: Implements a programmatic verification workflow to cross-reference metadata across multiple academic sources before inclusion in a manuscript.
Audit Metadata