Execute seven-move operating loop

Execute multi-agent operating loop with feedback cycles

Closed-loop multi-agent systems that refine state iteratively until convergence or completion.

• · loop configuration (agents, iteration count or until condition)
• · initial state
• · optional: step delay, budget cap

• · per-iteration: agent outputs, state delta, convergence metrics
• · final state and convergence report
• · failure diagnostics (if diverged or timeout)

• · multi-agent execution
• · state management
• · convergence monitoring

• · Agent specs defined (inputs, outputs, schemas)
• · Initial state valid and complete
• · Convergence criteria well-defined or iteration limit set

• · Agent timeout mid-iteration → iteration marked failed, loop pauses
• · State corruption → state integrity check fails, loop halts
• · Convergence never reached → loop exits at iteration limit, state reported
• · Budget exceeded → loop halts, partial results available

• · Per-iteration convergence metrics surface progress
• · Budget cap prevents runaway loops
• · State integrity checked each iteration
• · Failure diagnostics enable debugging divergence