workflow-manager

Multi-agent development workflows are orchestrated by coordinating specialized AI agents like Claude, Cursor, and Gemini through structured planning, validation, implementation, verification, and completion phases. The workflow manager manages agent interactions, state persistence, and context versioning.

Ensuring code quality with multiple AI agents requires a structured workflow with built-in validation and verification phases. The process applies configurable approval policies and conflict resolution mechanisms to maintain architectural soundness across agent contributions.

Yes, SurrealDB is used for state persistence to track progress across the five development phases. It manages agent context versioning and coordinates interactions for features defined in PRODUCT.md.

To start implementing a feature, define the requirements in PRODUCT.md and invoke the workflow manager to initiate the process. The system then orchestrates specialized agents through planning, validation, implementation, verification, and completion phases.

Yes, the workflow manager coordinates specialized agents by integrating Claude, Cursor, and Gemini for distinct roles in the development lifecycle. It manages their interactions to ensure structured AI-assisted software development with built-in quality assurance.

The phase-based workflow requires features to be explicitly defined in PRODUCT.md and relies on SurrealDB for state persistence. Users must configure approval policies and manage context versioning to handle potential conflicts between specialized agents.