The Agent Skills Directory

[SAFE]: The skill implements a legitimate administrative workflow for managing and improving AI agents within a defined platform. It uses a structured loop (Pin, Survey, Basis, Range, Analyze, Propose, Apply) to ensure changes are grounded in evidence.
[SAFE]: All tool invocations (getAgent, listRuns, aop-writer, etc.) are standard operational tools for the platform. The documentation emphasizes user confirmation before any modifications are applied.
[SAFE]: The skill incorporates clear security and privacy practices, such as handling sensitive data silently and requiring explicit user authorization for production-impacting changes.
[SAFE]: References to 'duvoai' and related domains/tools are consistent with the identified trusted vendor profile and represent the legitimate infrastructure of the skill's author.

improve-agent