When no data was returned for ignored tests, the LLM swarm inferred failure and shifted hypotheses, prompting the team to create synthetic responses.
