implement-paper-auto
Pass
Audited by Gen Agent Trust Hub on Mar 26, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill uses
curlto fetch paper overviews and full text fromhttps://alphaxiv.org. This involves downloading content from a third-party service that is not on the trusted vendor list, which then directly influences automated code generation.- [COMMAND_EXECUTION]: Instructions specify executingcurlcommands using a{PAPER_ID}variable derived from user-provided input. This creates a surface for command injection if the agent fails to strictly parse or sanitize the ID before passing it to the shell.- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8) by fetching untrusted markdown data from AlphaXiv to generate executable Python code. \n - Ingestion points: Markdown content fetched via
curlfrom AlphaXiv as defined inSKILL.md.\n - Boundary markers: The instructions do not define boundary markers or 'ignore' directives to prevent instructions within the paper text from being followed by the agent.\n
- Capability inventory: The agent is empowered to generate Python code for Marimo notebooks and create custom JavaScript/Python widgets using the
anywidgetframework.\n - Sanitization: There are no guidelines provided for validating or sanitizing the external markdown content before it is processed by the LLM for code generation.
Audit Metadata