multi-model

You can recover a stalled AI subagent by triggering a fresh-context retry to clear misleading state history, or by rotating to a different model from the registry to bypass model-specific execution bottlenecks.

Your LLM subagent may fail to progress due to accumulated context errors or model-specific limitations in reasoning. This Skill implements a recovery strategy using fresh-context resets and model-based retries to resolve these execution stalls.

You can automate model switching by integrating this Skill with the agent registry and subagent calling interface. It dynamically adjusts execution parameters and rotates to a different model when a subagent encounters model-specific performance regressions.

Yes, you can use context resets to fix orchestration bottlenecks. This Skill spawns a new instance of the same agent to clear misleading state history, facilitating complex task orchestration when agents encounter execution bottlenecks.

The best way to retry failed subagent tasks is to use a model-switching fallback. After a fresh-context retry fails, this Skill automatically switches to a more capable model from the registry to leverage varying reasoning strengths.

Yes, you need an agent registry and subagent calling interface to use the multi-model recovery strategy. These dependencies are required to dynamically adjust execution parameters and manage context resets for stalled agents.