A user on Reddit has reported a significant bug within Anthropic's platform, specifically concerning the management of sub-agent templates. The user claims that Anthropic's system ignores user-defined instructions for model and thinking levels when creating templates for sub-agents. This issue is particularly problematic as it appears to override user choices, leading to unexpected model behavior, such as Opus 5 being used when Sonnet 5 was explicitly selected. The user suggests this bug could be a contributing factor to high sub-agent usage costs. AI
IMPACT Potential issues with model selection and cost management for users employing sub-agents.
RANK_REASON User-reported bug in a platform's functionality.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
