The Agent Skills Directory

[SAFE]: The skill's primary function is to enforce accuracy and prevent hallucinations by requiring the agent to ground its responses in verifiable documentation or tool output.- [SAFE]: No obfuscation, data exfiltration, or unauthorized command execution patterns were found. All instructions align with safe and responsible agent behavior.- [INDIRECT_PROMPT_INJECTION]: The skill defines a protocol for ingesting data from external sources such as official documentation sites and tool help outputs (Ingestion points). It recommends using markdown comments to record sources (Boundary markers) and utilizes tool execution or web fetching (Capability inventory) to retrieve these facts. While no specific data sanitization is mentioned, the skill explicitly prioritizes official sources over third-party content, which is a significant mitigation factor for this surface area.

grounding