skills/mukul975/anthropic-cybersecurity-skills/performing-firmware-extraction-with-binwalk/Gen Agent Trust Hub
performing-firmware-extraction-with-binwalk
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The Python helper script scripts/agent.py uses the subprocess module to run binwalk commands for signature scanning and entropy analysis.\n
- Evidence: The script calls subprocess.run() with list-based arguments in functions such as run_binwalk_scan and run_binwalk_extract.\n
- Context: This implementation is safe because it does not use a shell environment for execution, mitigating shell injection risks.\n- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by reading raw binary data from potentially untrusted firmware images and outputting decoded strings to the agent's context.\n
- Ingestion points: Binary data is read from firmware files in the scan_strings function within scripts/agent.py.\n
- Boundary markers: No delimiters are present to separate untrusted data output from the agent's core instructions.\n
- Capability inventory: The skill has the capacity to read local files and execute binary analysis utilities.\n
- Sanitization: Bytes are decoded to ASCII with replacement for non-printable characters, but the content is not filtered for adversarial instruction patterns.
Audit Metadata