Current focus:
- building and refining M8Shift;
- designing role, relation, permission, and validation models for AI agents;
- writing content for a personal CV / builder profile site;
- exploring AstroPaper as a clean base for a personal website;
- improving repository-first AI workflows;
- turning repeated copy-paste between agents into structured handoffs;
- keeping the human decision point explicit.
M8Shift status
M8Shift is currently an active prototype.
The first beta 1.x prototype avoided 22 manual copy-paste exchanges during an early workflow.
By version 3.9, it supported around 80 agent exchanges over nearly 8 hours of work.
These are not lab benchmarks. They are field observations from real usage.
The lesson is simple: once several AI agents are involved, manual handoff becomes the bottleneck.