skills/yy/claude-scholar/check-refs/Gen Agent Trust Hub

check-refs

Pass

Audited by Gen Agent Trust Hub on May 6, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install and execute 'bibsleuth', an unverified external tool, using 'uv tool install' and 'uvx'.
  • [COMMAND_EXECUTION]: Uses shell commands to manage tool installation, run the bibliography checker, and interact with the filesystem.
  • [PROMPT_INJECTION]: An indirect prompt injection surface is present as the skill processes external bibliography data and summarizes tool output.
  • Ingestion points: Reads content from user-provided '.tex' and '.bib' files, and processes reports from 'bibsleuth-report.md'.
  • Boundary markers: No specific delimiters or warnings are provided to the agent to ignore potential instructions embedded in the processed LaTeX or report files.
  • Capability inventory: The skill possesses the ability to execute shell commands ('uvx'), read sensitive project files, and modify the bibliography file ('.bib').
  • Sanitization: There is no evidence of sanitization or validation of external content before it is incorporated into the agent's context for summarization.
Audit Metadata
Risk Level
SAFE
Analyzed
May 6, 2026, 02:29 AM
Security Audit — agent-trust-hub — check-refs