TraceLab: Characterizing Coding Agent Workloads for LLM Serving

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Kan Zhu, Mathew Jacob, Chenxi Ma, Yi Pan, Stephanie Wang, Arvind Krishnamurthy, Baris Kasikci
• Venue: arXiv: cs.LG
• Published: 2026-06-29
• Collected: 2026-06-30
• URL: http://arxiv.org/abs/2606.30560v1
• PDF: https://arxiv.org/pdf/2606.30560v1
• DOI: UNKNOWN

Plain-Language Brief

Researchers present TraceLab, a dataset of 4,300 coding-agent sessions collected from their own usage of Claude Code and Codex. The data reveals characteristics of these workloads, such as long autonomous loops, long contexts with short outputs, diverse tool calls, and high but imperfect prefix cache hit rates. These findings suggest potential optimization strategies for serving, like lower-overhead tool calling, append-length-aware prefill, semantic-aware tool-latency prediction, and improved KV-cache management.

What the Document Says

• The researchers present TraceLab, a dataset containing approximately 4,300 coding-agent sessions. The data is from their personal usage of Claude Code and Codex.
• The analysis of the data shows that coding-agent workloads consist of long autonomous loops, long contexts with short outputs, diverse and heavily-tailed tool calls, and high but imperfect prefix cache hit rates.

Key Concepts Mentioned

• coding agents
• LLM serving
• data analysis
• workload patterns

Claims Made by the Document

• Coding-agent workloads feature long autonomous loops.
• Coding-agent workloads consist of long contexts with short outputs.

Evidence Used

• '...long autonomous loops...' (indirect evidence from the statement 'our analysis shows that coding-agent workloads feature...')
• '...long contexts with short outputs...' (indirect evidence from the statement 'our analysis shows that coding-agent workloads consist of...')

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Coding agents are rapidly becoming a major application of agentic LLMs, but serving them efficiently remains challenging. Progress on this challenge requires understanding real workload patterns, yet the data needed for such analysis is largely absent. Existing public traces and benchmarks do not capture real, day-to-day coding-agent usage across multiple agents and model families for serving-system analysis. To help fill this gap, we collect and release a trace of roughly 4,300 coding-agent sessions, containing about 350,000 LLM steps and 430,000 tool calls from our own day-to-day use of Claude Code and Codex. Our analysis shows that coding-agent workloads feature long autonomous loops, long contexts with short outputs, diverse and heavily-tailed tool calls, and high but imperfect prefix cache hit rates. These findings point to concrete opportunities for optimizing serving, including lower-overhead tool calling, append-length-aware prefill, semantic-aware tool-latency prediction, and improved KV-cache management around human-paced gaps. We release the dataset, trace collection pipeline, and analysis code at https://github.com/uw-syfi/TraceLab.git; the project website is https://tracelab.cs.washington.edu.