Expired50% confidenceFactExact time
In agent scenarios, due to multiple rounds of tool calls, each call must carry the previous conversation history and system prompt, causing input tokens to grow cumulatively, and a code refactoring task may consume tens of thousands or even hundreds of thousands of tokens.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 9/30/2026(expired)
Sources
Observe Every Step of Claude Code Execution with Eflow Base: Full Tracing Configuration Guide
bilibili老文-taichuy6/6/2026
Related Claims
UnverifiedLong conversations package and resend the entire context history to the model, which is counted toward the token usage of each request; the longer the conversation, the more tokens a single request consumes.76% similarUnverified传统Workflow Agent模式下所有任务处理逻辑需以System Prompt形式预先注入大模型上下文窗口，提示词可能膨胀到数万甚至数十万Token73% similarUnverifiedIn programming assistant scenarios, system prompts and code context typically consume a large number of tokens, resulting in an extremely high cache hit rate.72% similarUnverifiedIn multi-turn conversations, all previous context is resent as input each time, so redundant code generated early on continues to incur token costs in every subsequent round.71% similarUnverified程序化工具调用相比传统 ReAct 框架能降低 Agentic 工作流的 Token 消耗和延迟，在串联数十次工具调用的复杂任务中效率提升显著71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/51967API
curl https://kongchang.com/api/v1/knowledge/claims/51967MCP
get_claim(id=51967)