radare2

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill implements an on-demand bootstrap system that downloads radare2 binaries from the official GitHub repository (radareorg/radare2) if they are not already installed on the system.
  • [COMMAND_EXECUTION]: The skill executes local radare2 utilities (r2, rabin2, etc.) to analyze binaries. The provided scripts correctly use the -- argument separator to ensure user-supplied file paths are treated as targets rather than command flags, preventing command injection.
  • [DATA_EXPOSURE]: The skill is designed to extract metadata, strings, and header information from binary files. While this involves reading user-provided files, it is the intended and essential functionality of a reverse engineering tool.
  • [PROMPT_INJECTION]: The skill has a potential surface for indirect prompt injection because it processes untrusted binary data (ingestion points: $TargetPath in recon.ps1 and recon.sh). Maliciously crafted strings or function names within a binary could attempt to influence agent behavior. However, this is a characteristic risk of analysis tools and is considered low impact within its intended use case.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 01:33 AM
Security Audit — agent-trust-hub — radare2