Persistent Context Graphs for Efficient Memory Compaction in LLM Agents

arXiv:2609.40118 · cs.CL · Submitted 2026-09-30 · Read on arXiv

cs.CL

Submitted: 2026-09-30

Updated: 2026-09-30

Code: https://github.com/UCSB-NLP-Chang/ReCAP

Terminology

• KV cache — 17× in this paper · explained in 3 episodes
• Accuracy — 10× in this paper · explained in 1 episode
• Core — 4× in this paper · explained in 2 episodes
• Benchmark — 3× in this paper · explained in 2 episodes
• Trace — 3× in this paper · explained in 2 episodes
• Ablation — 2× in this paper · explained in 1 episode
• Adam — 2× in this paper · explained in 2 episodes
• Ablation Studies — 1× in this paper · explained in 1 episode
• Agent Harness — 1× in this paper · explained in 2 episodes
• Agent Memory — 1× in this paper · explained in 1 episode
• Distillation — 1× in this paper · explained in 2 episodes
• Embeddings — 1× in this paper · explained in 2 episodes
• Graph Construction — 1× in this paper · explained in 3 episodes
• Pruning — 1× in this paper · explained in 4 episodes
• Sparsity — 1× in this paper · explained in 4 episodes
• Transformer — 1× in this paper · explained in 2 episodes

Sources

• ACON: Optimizing Context Compression for Long-horizon LLM Agents
• IntentKV: Cross-Turn Intent-Aware KV Cache Pruning for Agent Inference
• The Complexity Trap: Simple Observation Masking Is as Efficient as LLM Summarization for Agent Context Management
• Practical Online KV Cache Compaction for LLM Agents: An Empirical Study
• Mix-Quant: Quantized Prefilling, Precise Decoding for Agentic LLMs
• gpt-oss-120b & gpt-oss-20b Model Card
• Scaling Long-Horizon LLM Agent via Context-Folding
• Quest: Query-Aware Sparsity for Efficient Long-Context LLM Inference
• ContextWeaver: Selective and Dependency-Structured Memory Construction for LLM Agents
• SWE-Together: Evaluating Coding Agents in Interactive User Sessions
• Qwen3 Technical Report
• AttnCompress: Dynamic Attention-Guided Trajectory Compression for Software Engineering Agents
• TraceLab: Characterizing Coding Agent Workloads for LLM Serving
• Fast KV Compaction via Attention Matching

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering