The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Bash and Python to execute a wide array of forensics tools such as Volatility 3, Tshark, Binwalk, and TestDisk. These operations are standard for digital forensics tasks involving the inspection of system artifacts.
[EXTERNAL_DOWNLOADS]: Recommends the installation of numerous well-known security and forensics tools via system package managers (apt, brew) and language-specific registries (pip, gem). It also references specific utility repositories on GitHub for tasks like KeePass database cracking and Git repository recovery.
[CREDENTIALS_UNSAFE]: Features extensive guidance on recovering and cracking credentials from various sources, such as Windows SAM databases, KeePass files, and browser data stores. These features are documented as standard forensics procedures.
[DATA_EXFILTRATION]: Documents methods for identifying and reconstructing data that has been exfiltrated through covert network channels, such as DNS query names, ICMP payloads, and TCP flags.
[PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it is designed to ingest and analyze untrusted data from external sources (forensic artifacts) that could contain malicious instructions.
Ingestion points: Processing of server logs, disk images, memory dumps, and packet captures as described in SKILL.md, network.md, and disk-and-memory.md.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are used when reading data from untrusted files.
Capability inventory: The agent has access to Bash (shell), Python (scripting), and WebFetch (network), which could be leveraged if an injection is successful.
Sanitization: Input from forensic artifacts is not sanitized before being presented to the agent for analysis.

ctf-forensics