Research existing Drupal workflow execution models for suitability in agentic workflow generation

FlowDrop aims to provide a visual, agentic workflow execution engine capable of orchestrating complex tasks, including AI-driven decision-making and multi-step automations. Before committing to a fully custom execution model, we need to evaluate whether existing Drupal core or contrib workflow systems can serve as a foundation or partially integrate into FlowDrop.

Scope of Research:

• Drupal Core Workflow / Workbench:
  • Entity-based state management
  • Transition handling and limitations for dynamic or AI-driven flows
• ECA (Event-Condition-Action) module:
  • Node-based event and condition orchestration
  • Suitability for high-volume, asynchronous processing and complex branching
• Core Actions and Events:
  • Trigger-based execution model
  • Extensibility for AI-driven or agentic processes
• Investigate any other contrib solutions relevant to workflow orchestration

Goals:

• Identify gaps in current contrib solutions for supporting FlowDrop’s complex workflow needs (dynamic branching, agentic reasoning, async execution, external API orchestration).
• Determine if we should extend an existing model (ECA or Core Workflow) or implement a custom Node Runtime as designed in FlowDrop.
• Produce a recommendation document that guides architectural decisions for the FlowDrop execution engine.

Deliverables:

• Comparative matrix highlighting features, limitations, and performance considerations of each evaluated model
• Recommendation on integration, extension, or custom build approach