The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents the use of standard security tools such as binwalk, ripgrep (rg), hexdump, strings, readelf, and qemu to analyze firmware components. These commands are executed by the user as part of a manual analysis workflow.
[PRIVILEGE_ESCALATION]: The skill mentions the use of sudo for commands like mount, chroot, and running analysis frameworks such as firmae and emba. These operations are typical for firmware analysis (e.g., mounting extracted filesystems) and are presented as part of legitimate administrative tasks.
[INDIRECT_PROMPT_INJECTION]: The skill describes a process that involves ingesting and analyzing untrusted external data (firmware images). While this creates an attack surface where an agent could process malicious instructions embedded in a target's configuration files or binary strings, the skill follows standard security research practices and does not include any specific exploitable patterns targeting the agent itself.
[SAFE]: The overall instructions are consistent with established security auditing methodologies. No hidden payloads, obfuscated code, or suspicious network connections were identified.

analyzing-firmware-images