latex-diff

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands like git show <REV_OLD>:<path> and git worktree add /tmp/ld-old <REV_OLD>. These templates use user-supplied revisions and paths directly, which could allow for command injection if malicious strings containing shell metacharacters are provided.
  • [PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by ingesting and summarizing untrusted LaTeX document content.
  • Ingestion points: The OLD and NEW LaTeX files or project directories identified in the Protocol section of SKILL.md.
  • Boundary markers: Absent. The skill does not use specific delimiters or instructions to disregard embedded text commands.
  • Capability inventory: The skill is authorized to use Bash for git, latexdiff-agent, and file system operations (mktemp, mkdir, rm), providing a significant capability surface for any successful injection.
  • Sanitization: Absent. There is no mention of escaping or validating the content of the LaTeX files or the user-provided revision identifiers.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 06:17 AM
Security Audit — agent-trust-hub — latex-diff