[Suggestion] Consider methodology suggested in RollingDocs: A Framework for Human-AI Collaborative Development

[vcnd-recharged]

Hey guys, i wrote this orchestration framework some time ago. It looks like it matches the methodology you put into the spec-kit. If you curious, please give that a review and let me know what you think!

https://github.com/Viktor286/rolling-docs

RollingDocs is an experimental, role-based, document-driven, Human-AI team & collaborative system of development orchestration.

On a very high-level, it's a combination of Project/Team Management where everything is a document, everything is a prompt.

This framework explores a structured approach to managing software development projects where contributions may come from both Human experts and specialized AI models. It utilizes a system of version-controlled markdown documents stored directly within the project repository to serve as a shared knowledge base, communication protocol, and operational guide.

Key Concepts of RollingDocs

• Role-Based: Work is defined by abstract Roles (e.g., Developer, QA, PM, Framework Guardian), each with documented responsibilities (Description.md) and workflows (Instructions.md).
• Document-Driven: All necessary context – project goals, technical plans, standards, tooling, task specifics – resides in version-controlled markdown files within the RollingDocs/ directory.
• Human-AI Interchangeability: By defining roles via process and documentation interfaces, any capable contributor (Human or AI) can fulfill a role.
• Task Management via File System: Individual work units (Task Specifications) are managed as markdown files moved through status folders (RollingDocs/Tasks/Planned/, Progress/, Completed/).
• Integrated QA: Quality assurance checks are a mandatory, defined step in relevant workflows.
• Framework Guardian: A dedicated role maintains the integrity and evolution of the RollingDocs framework itself.
• AI-Specific Directives: Includes a convention ([AI: ...]) for providing explicit, mandatory instructions to AI contributors within the documentation.