process

To manage AI-assisted development workflows, you use a lean process framework that separates workspace and per-project contexts, enforcing a repeatable session lifecycle and cross-project consistency for human-AI teams.

The best way to track design decisions in a human-AI coding session is using a decisions.md file with a lifecycle that moves entries from active to settled to demonstrated, preserving rationale and enabling Foundry annotations.

You start and stop a collaborative coding session by issuing the /start [project] command to bootstrap and load process files, then concluding work with the /stop command to ensure proper lifecycle closure and quality gate enforcement.

You need process.md, next.md, and decisions.md in your workspace to bootstrap a lean development session, along with an optional CLAUDE.md file to support session initialization and load relevant project context.

You can use this lean process framework for multi-project AI work because it explicitly defines workspace versus project scope, providing Foundry conventions, quality gates, and repeatable session lifecycle commands to support diverse environments.