dsh-skill-autopilot: Review skill evolution suggestions using a model and access via an anti-fragmentation strategy.

dsh-skill-autopilot is a DSH workflow plugin maintained by BrainZombie. It submits pending suggestions for Mnemon skill evolution to a configured Large Language Model (LLM) for review, passing them through an anti-fragmentation deterministic gate for adjudication. Upon approval, they are packaged into a SKILL.md and persisted to the skill data source, including version history and rollback capabilities. The article details the five stages of a single run, anti-fragmentation admission rules, packaging linting, PowerShell installation commands, slash commands and agent tool invocation, decision log location, configuration items, testing methods, as well as permissions and autoApply usage considerations.