The Agent Skills Directory

[SAFE]: The skill defines a workflow where the agent acts on instructions provided by an external MCP server (the next_action field). To mitigate risks associated with indirect prompt injection or malicious server responses, the skill includes explicit validation requirements for the agent to verify the safety and scope of these directives before proceeding.
[SAFE]: Instructions are provided to ensure no raw API keys, tokens, or credentials are included in query strings or payloads, preventing accidental data exposure.
[SAFE]: The skill explicitly defers to host-level security policies and permission models, ensuring that the automation does not attempt to bypass existing environment constraints.

kairos