guild-loop-implement

To validate plans against challengers, the skill coordinates multi-layer adversarial loops across a lane's delivery lifecycle, managing L3, L4, and security-review interactions with per-lane isolation and structured audit trails.

Adversarial loop execution manages restarts via deterministic restart logic and per-lane counters, providing handoff receipts and emitting structured JSONL events to maintain a reliable audit trail across restarts.

Adversarial validation loops are triggered when a lane activates the loops feature via --loops flags set to implementation or all, applying plan validation constraints and delegating interactions to specialized challengers.

The lane workflow delegates nested dialogues, sentinel termination detection, and cap-aware progress management to the execute-plan component, ensuring structured plan validation and escalation prompts across isolated lanes.

Running multi-layer adversarial loops requires explicit activation via --loops flags and depends on specialized challengers, meaning loops will not execute automatically without proper lane configuration and per-lane isolation setup.