---
格式版本: 2
标题: "Token Optimization and Context Window Management in Multi-Agent AI Workflows"
原文链接: "https://arxiv.org/abs/2608.17188"
发布日期: "2026-08-17"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Mon, 17 Aug 2026 22:56:50 UTC (702 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-20T15:41:41+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-20T15:33:55+08:00"
入库时间: "2026-08-20T07:41:41.634Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=latency&searchtype=all"
匹配关键词:
  - "latency"
  - "AI"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 15
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "论文主题为多智能体AI工作流中的token优化与上下文管理，完全不涉及超节点、AI Rack、机柜级AI基础设施、互连、供电、液冷等任何项目主题，仅命中通用词latency，明显无关。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-20T15:47:10+08:00"
AI主题相关性: 0
AI来源权威性: 5
AI新颖性: 5
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 5
AI摘要: "论文提出多智能体AI工作流中token优化与上下文窗口管理的实践框架，涵盖六种可复用模式。生产环境中，该方法将冷加载延迟从约数分钟降至61-116秒，并减少60-70%的token消耗；"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-07T03:25:06.319Z"
采集批次: "2026年8月20日14点19分32秒"
采集批次ID: "20260820-141932-079"
去重键: "https://arxiv.org/abs/2608.17188"
---

## Computer Science > Computation and Language

## Title:Token Optimization and Context Window Management in Multi-Agent AI Workflows

Authors:[Dvir Shamay](https://arxiv.org/search/cs?searchtype=author&query=Shamay,+D)

[View PDF](https://arxiv.org/pdf/2608.17188)

> Abstract:Multi-agent AI workflows are limited not only by model quality but by token cost, latency, and context-window quality. This paper presents a practitioner framework for token optimization and context-window management, grounded in an internal production dashboard that extracts structured work items from meetings, email, and chat with LLMs and routes summaries across workstreams. Six patterns are described: context stratification, fetch-once/process-locally architecture, schema-contracted prompts, token-aware fallback chains, semantic caching, and inter-agent communication compression. In production they cut measured cold-load latency to 61-116 seconds (six timed runs) from an operational baseline of roughly 3.5-10.5 minutes, with an estimated 60-70% token reduction. It also reports a controlled context-composition study: 2,420 confirmatory trials across 11 model configurations, using 661 anonymized workplace items scored for relevance. Holding the prompt at a fixed ten items, replacing some high-relevance items with same-domain low-relevance items improves the model's relevance-score concordance on the target items, versus high-relevance items only; we call this relevance-contrast context. In the all-11 paired analysis, the 50:50 signal/noise condition improved relevance accuracy by +0.077 over the 100% condition (naive 95% CI \[+0.056, +0.098\], Cohen's d = 0.49, Holm-adjusted p <.001, n = 220). These cells are not independent; by the nine model families the effect is +0.084 (95% interval \[+0.064, +0.103\]), reported as a within-corpus descriptive comparison, not a population inference. A Fusion-of-N follow-up found that learned synthesis did not beat the mechanical set union of item IDs. The contribution is a measured engineering layer between model research and production agent practice: repeatable patterns and evaluation methods for faster, cheaper, more reliable workflows.

| Comments: |  |
| --- | --- |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | [arXiv:2608.17188](https://arxiv.org/abs/2608.17188) \[cs.CL\] |
|  | (or [arXiv:2608.17188v1](https://arxiv.org/abs/2608.17188v1) \[cs.CL\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.17188](https://doi.org/10.48550/arXiv.2608.17188) |

## Submission history

From: Dvir Shamay \[[view email](https://arxiv.org/show-email/2a0af114/2608.17188)\]  
**\[v1\]** Mon, 17 Aug 2026 22:56:50 UTC (702 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.17188) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))