re-fw-extract
Warn
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONOBFUSCATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to download source code for the sasquatch utility from a third-party GitHub repository (https://github.com/devttys0/sasquatch) which is not in the trusted vendor list.
- [COMMAND_EXECUTION]: The analysis workflow involves using package managers like apt and pip, as well as compiling software using make, which involves executing system-level commands.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze untrusted firmware images using tools like strings and binwalk. This creates an attack surface where malicious instructions embedded in the firmware could be interpreted by the agent.
- Ingestion points: firmware.bin processed by strings, binwalk, and hexdump.
- Boundary markers: None identified.
- Capability inventory: Package management, git, and build tools.
- Sanitization: No validation of firmware content is performed.
- [OBFUSCATION]: The content includes hex-encoded byte sequences (e.g., \x03\x04) for binary signature identification. While functionally necessary, these are detected as obfuscated character escapes.
Audit Metadata