The Agent Skills Directory

[SAFE]: The skill is purely instructional and logic-based. It defines a protocol for an AI agent to decide whether and how to ask questions to a user. No network operations, file system modifications, or sensitive data access were detected.
[SAFE]: The skill uses structured YAML and YON files to define its logic. The execution flow focuses on internal triage, classification of stakes, and quality-gating questions using other local skills like /insight-critique and /insight-assess.
[SAFE]: There are no external downloads or unverifiable dependencies. All mentioned successor skills (insight-cross-examine, insight-assess, reflect) and sibling protocols (caller-options, next-skills) are relative references to what appears to be a local ecosystem of skills.
[SAFE]: No prompt injection, obfuscation, or data exfiltration patterns were found. The 'Self-improvement' step follows safe practices by requiring user approval for any proposed changes and explicitly forbidding auto-application.

ask-gate