The Agent Skills Directory

[SAFE]: The skill is designed as an advisory and ideation tool. It includes a specific rule rule:no-execution which strictly forbids the agent from making changes, writing code, or executing commands.
[SAFE]: The sub-agent orchestration mechanism described as 'cold fan-out' utilizes platform-native isolated workers for independent analysis. The protocol ensures these are 'leaf' agents that cannot spawn further sub-skills, preventing uncontrolled execution depth.
[SAFE]: No sensitive file access, network operations, or external dependencies were found. The skill operates within the provided context to generate analytical insights.
[SAFE]: Self-improvement mechanisms are gated behind human-in-the-loop (HIL) approval, ensuring that any proposed updates to the internal lens roster must be reviewed before application.

insight-angles