Expired95% confidenceFactTime unknown
The MARVIS team performed no task-related training or fine-tuning on the model, assigning each Agent its role definition solely through system prompts.
1
Sources
95%
Confidence
Medium-term (~90 days)
Relevance
6/1/2026
First Seen
Valid until: 8/30/2026(expired)
Sources
Related Entities
Related Claims
UnverifiedClaude Code 的 Team、Enterprise 和 API 用户默认不会被用于模型训练64% similarUnverifiedSystem Prompt与用户输入本质上都是模型的输入token，模型并不存在独立的规则执行引擎来强制保证指令被遵守56% similarUnverified通过Bedrock访问Claude模型，用户无需直接与Anthropic签约，且数据不会被用于模型训练54% similarUnverifiedThe Orchestrator does not create new Agents; it only orchestrates existing Agents, and does not write code or run tests itself.54% similarUnverifiedDeepSeek V4 Pro did not provide a To-Do list for task decomposition when executing tasks.52% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/21609API
curl https://kongchang.com/api/v1/knowledge/claims/21609MCP
get_claim(id=21609)