The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/agent.py script utilizes the subprocess.run module to interact with the Windows schtasks utility. The script calls schtasks /query to list tasks and export their XML configurations. The usage is implemented using best practices by passing arguments as a list rather than a shell string, which prevents shell injection vulnerabilities. This is an expected and legitimate function of a threat hunting skill.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from system logs and task definitions which could contain malicious content designed to influence an AI agent.
Ingestion points: System data is ingested from schtasks output in scripts/agent.py and from local log files (JSON/CSV) in scripts/process.py.
Boundary markers: The skill does not employ explicit delimiters or system instructions to distinguish between its own logic and the data ingested from external system logs.
Capability inventory: The skill has the capability to execute OS commands (schtasks) and write analysis results to the local filesystem.
Sanitization: The scripts focus on identifying indicators of compromise using regular expressions but do not perform formal sanitization of the raw data before it is processed or reported.

hunting-for-scheduled-task-persistence