comm-lit-review-claude-single

Pass

Audited by Gen Agent Trust Hub on Mar 18, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection attacks.
  • Ingestion points: The skill ingests untrusted data by reading local PDFs from 'papers/' and 'literature/' directories and fetching content from external sites like IEEE Xplore, ScienceDirect, and ACM Digital Library (SKILL.md).
  • Boundary markers: Absent. The instructions do not include delimiters or warnings to ignore instructions embedded within the research material.
  • Capability inventory: The skill allows access to 'Bash(*)', 'Read', 'Write', and 'WebFetch' (SKILL.md).
  • Sanitization: Absent. There is no evidence of filtering or sanitization of the content retrieved from external sources.
  • [COMMAND_EXECUTION]: The skill enables the 'Bash(*)' tool, which provides the agent with extensive command-line capabilities. While this is intended for file management and processing, the broad scope of the tool increases the potential impact if the agent is manipulated via malicious content in a paper.
  • [EXTERNAL_DOWNLOADS]: The skill fetches content from well-known academic services including 'ieeexplore.ieee.org', 'sciencedirect.com', and 'dl.acm.org' to provide research summaries.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 18, 2026, 04:29 PM