AI collaboration isn’t fully captured by final output alone. The important part is the path: questions, answers, tradeoffs, and decisions accumulated along the way.
pmpt records that decision flow so the development journey becomes reproducible and shareable.
It manages two documents:
- `pmpt.ai.md` for AI-facing instructions
- `pmpt.md` for human-facing progress and decision logs
The usage flow is simple:
`pmpt init -> pmpt plan -> pmpt save` (and `publish` / `clone` when needed).
You can explore published projects and `clone` them to absorb how others approached similar problems, then continue from there in your own direction.
One open problem I’m still working on is how to keep AI consistently maintaining these records over time. MCP support helps, but I’m still iterating on the right balance.
AI collaboration isn’t fully captured by final output alone. The important part is the path: questions, answers, tradeoffs, and decisions accumulated along the way.
pmpt records that decision flow so the development journey becomes reproducible and shareable.
It manages two documents:
- `pmpt.ai.md` for AI-facing instructions - `pmpt.md` for human-facing progress and decision logs
The usage flow is simple: `pmpt init -> pmpt plan -> pmpt save` (and `publish` / `clone` when needed).
You can explore published projects and `clone` them to absorb how others approached similar problems, then continue from there in your own direction.
One open problem I’m still working on is how to keep AI consistently maintaining these records over time. MCP support helps, but I’m still iterating on the right balance.