dual-loop

Agent orchestration coordinates task delegation by issuing a Strategy Packet from a strategic Outer Loop to a tactical Inner Loop, executing the task, and verifying the delta outcomes. It separates planning from execution to ensure reliable iterative correction.

A dual-loop agent architecture separates strategic planning from tactical execution to enable iterative task verification. You need it when decomposing complex tasks requires handing off goals to an executor and formally correcting the structured outcomes.

To generate a strategy packet, define your goals, create a Pre-Execution Diagram, specify allowed file paths, and set strict non-GIT constraints for the inner loop. The packet is then issued to the tactical executor to guide execution and reporting.

Yes, this agent orchestration workflow applies across framework-agnostic contexts. It relies on formal strategy packets and structured correction reports rather than specific frameworks, ensuring task decomposition and delta verification work in any environment.

Delta verification works by logging the outcomes of the tactical Inner Loop and comparing them against the strategy packet goals. If the results do not match, structured correction workflows are applied to iteratively converge on the correct outcome.

Strict no GIT constraints prevent the tactical inner loop from making unauthorized version control changes during execution. This ensures the executor operates only within allowed file paths, keeping the workflow safe for delta verification and iterative correction.