The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill monitors file system activity and parses logs, creating an attack surface for indirect prompt injection.
Ingestion points: The scripts scripts/agent.py and scripts/process.py read file names, metadata, and log entries (e.g., in functions like analyze_honeypot_logs and detect_ransomware_indicators).
Boundary markers: There are no explicit delimiters or instructions for the agent to ignore potentially malicious content within the monitored data.
Capability inventory: The skill includes capabilities for file system modification and network operations (demonstrated via requests in the SIEM integration snippets).
Sanitization: The skill does not implement explicit sanitization of the data ingested from the file system or logs before processing it.
[COMMAND_EXECUTION]: The skill utilizes administrative shell commands to configure system security and network shares.
Evidence: SKILL.md includes PowerShell commands such as Install-WindowsFeature, New-SmbShare, and auditpol for managing Windows File Server Resource Manager and auditing policies.
[EXTERNAL_DOWNLOADS]: The skill fetches document tokens from an external security service.
Evidence: SKILL.md contains curl commands targeting canary.tools to generate and download canary files used for ransomware detection.

implementing-honeypot-for-ransomware-detection