The Agent Skills Directory

[SAFE]: The skill implements a structured interviewing process to help users build automations. It reads from local context files (priorities.md, about-me.md, connections.md) to provide relevant advice, which is standard behavior for personalized AI assistants. It only writes to a decision log and creates new skill files, following the principle of least privilege by remaining read-only for other existing user files. The generated artifacts include safety headers (bike-method-phase: 1) that prevent the autonomous execution of new tools without manual human validation, mitigating risks associated with automated code generation. No network exfiltration, obfuscation, or privilege escalation patterns were detected.

level-up