The Agent Skills Directory

[SAFE]: The skill implements legitimate cybersecurity workflows for industrial control systems. The Python script scripts/agent.py and the code blocks in SKILL.md perform local data processing on asset inventories and patch definitions to calculate risk scores and compliance status without any suspicious network or file system operations.
[EXTERNAL_DOWNLOADS]: The skill references official sources for vulnerability data, such as CISA (cisa.gov) and NIST (nist.gov). These are well-known, trusted government services used for legitimate security research and advisory tracking.
[COMMAND_EXECUTION]: Provides example curl and jq commands for querying vulnerability feeds and powershell snippets for WSUS management. These are documented as educational examples for the user/agent to follow for asset management and do not contain malicious payloads.
[INDIRECT_PROMPT_INJECTION]: The skill ingests JSON data (assets and patches) for processing. While this is an attack surface, the processing logic is strictly mathematical/logical (calculating dates and scores) and does not interpolate untrusted data into instructions that would influence the agent's behavior.

implementing-patch-management-for-ot-systems