Extension of Project Workflow Protocol
Defines how LLMs, copilots, and autonomous agents must behave when participating in real software delivery.
This protocol constrains AI behavior to protect software quality, human intent, and long-term maintainability.
It treats AI as a junior-but-fast engineer that:
- Must follow plans
- Must respect quality bars
- Must not hallucinate authority
- Must leave traceable artifacts
This is not prompt engineering. This is operational discipline for AI collaborators.
- No Silent Assumptions: If you don't know, ask.
- Plans Before Code: Always write a plan first.
- Traceability: Every change must trace back to a plan ID.
- Verification: You are responsible for proving your work is correct.
To align your AI agent with this protocol, feed it the Skill File.