The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands (e.g., mount, find, ls) to manage forensic evidence and traverse file systems. These operations are performed with read-only flags to preserve evidence integrity.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, untrusted data from forensic artifacts such as Windows Registry keys and XML task definitions. This represents an attack surface for indirect prompt injection if the artifacts contain malicious content designed to influence the agent.
Ingestion points: Extraction of registry values and scheduled task XMLs in SKILL.md and scripts/agent.py.
Boundary markers: Not present.
Capability inventory: Extensive file system access (read/write) and command execution capability.
Sanitization: Uses standard libraries like python-registry and ElementTree for artifact parsing.
[OBFUSCATION]: The script scripts/agent.py employs a simple string manipulation technique (.replace(". ", ".")) to define the path for SSH authorized keys. While minor, this tactic can be used to evade basic static analysis filters looking for sensitive file paths.
[CREDENTIALS_UNSAFE]: The skill is configured to access and display contents from sensitive locations such as SSH authorized_keys. Although this is a core requirement for malware persistence investigation, it involves handling sensitive credential-related data from the analyzed images.

performing-malware-persistence-investigation