The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous commands involving system-level tools such as mount, tshark, volatility, dd, kpartx, and nc. These are intended for data extraction and analysis within a forensic or CTF context.
[EXTERNAL_DOWNLOADS]: The documentation suggests cloning and building specialized forensic tools from non-verified GitHub repositories, such as ivanmrsulja/keepass2john and gtoubassi/femtozip. It also references installing various third-party Python packages like heatshrink2, regipy, and stego-lsb.
[PRIVILEGE_ESCALATION]: Instructions include the use of sudo for mounting disk images (mount -o loop,ro) and managing loop devices, which are necessary for forensic analysis but involve executing commands with elevated privileges.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data sources, including disk images, network captures (PCAP), and Windows event logs. It lacks explicit sanitization or boundary markers for the data handled by forensic tools, presenting a surface for indirect prompt injection attacks.
[CREDENTIALS_UNSAFE]: While primarily for forensic education, the skill contains hardcoded cryptographic keys and hex strings (e.g., in references/disk-advanced.md) as examples for specific challenges. It also documents techniques for extracting sensitive artifacts like the Windows SAM database and browser login data.

ctf-forensics