medical-imaging-review

Pass

Audited by Gen Agent Trust Hub on Mar 31, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: Provides detailed setup instructions and shell commands for executing third-party MCP servers (ArXiv, PubMed, and Zotero) from public GitHub repositories using the uvx tool.
  • [COMMAND_EXECUTION]: Instructs the use of the Bash tool to perform local API interactions with Zotero for managing bibliography data and extracting paper metadata.
  • [PROMPT_INJECTION]: The skill ingests untrusted text from external academic papers, which creates a surface for indirect prompt injection where malicious content in a paper could influence the agent's behavior.
  • Ingestion points: Paper content is retrieved via tools such as mcp__arxiv-mcp-server__read_paper and mcp__zotero__zotero_get_item_fulltext.
  • Boundary markers: The instructions do not define clear delimiters or use protective prompts to separate ingested literature from the agent's core instructions.
  • Capability inventory: The skill allows access to high-privilege tools including Bash, Write, and Edit.
  • Sanitization: No validation or sanitization of the retrieved scientific content is implemented before it is processed by the AI.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 31, 2026, 08:21 AM
Security Audit — agent-trust-hub — medical-imaging-review