performing-firmware-extraction-with-binwalk

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent or user to install several third-party tools and libraries required for analysis, including the binwalk3 Python package via pip, as well as system-level utilities like unsquashfs, jefferson, sasquatch, and firmware-mod-kit.
  • [COMMAND_EXECUTION]: The skill relies heavily on shell command execution to perform binary analysis (e.g., binwalk -B), entropy plotting (binwalk -E), and filesystem extraction. It also uses standard Linux utilities like strings, grep, and find to locate sensitive information within the extracted file structures.
  • [INDIRECT_PROMPT_INJECTION]: As the skill is designed to ingest and process untrusted external binary data (firmware images) and has significant capabilities (file system access and shell execution), it possesses a surface for indirect prompt injection. A malicious firmware image could theoretically contain crafted strings intended to influence the agent's behavior during the analysis phase. However, this is a risk inherent to the primary purpose of firmware analysis skills.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 10:26 PM
Security Audit — agent-trust-hub — performing-firmware-extraction-with-binwalk