# Learning When to Remember: Risk-Sensitive Contextual Bandits for Abstention-Aware Memory Retrieval in LLM-Based Coding Agents

arXiv: https://arxiv.org/abs/2604.27283 · HTML: https://mlsanity.com/paper/2604.27283/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Agent memory systems, one of 16 directions in LLM agents and tool use.
- Builds on 26 of its 31 references that are in the map.
- First posted on arXiv 30 Apr 2026.

## Placed in

- LLM agents and tool use / [Agent memory systems](https://mlsanity.com/agents/agent-memory-systems/)

## Builds on (most cited first, 26 of 26)

- [Memory-R1: Enhancing Large Language Model Agents to Manage and Utilize Memories via Reinforcement Learning](https://mlsanity.com/paper/2508.19828/) (2025)
- [MIRIX: Multi-Agent Memory System for LLM-Based Agents](https://mlsanity.com/paper/2507.07957/) (2025)
- [Agentic Memory: Learning Unified Long-Term and Short-Term Memory Management for Large Language Model Agents](https://mlsanity.com/paper/2601.01885/) (2026)
- [Process vs. Outcome Reward: Which is Better for Agentic RAG Reinforcement Learning](https://mlsanity.com/paper/2505.14069/) (2025)
- [Memory for Autonomous LLM Agents:Mechanisms, Evaluation, and Emerging Frontiers](https://mlsanity.com/paper/2603.07670/) (2026)
- [KnowledGPT: Enhancing Large Language Models with Retrieval and Storage Access on Knowledge Bases](https://mlsanity.com/paper/2308.11761/) (2023)
- [RAP-Gen: Retrieval-Augmented Patch Generation with CodeT5 for Automatic Program Repair](https://mlsanity.com/paper/2309.06057/) (2023)
- [Parametric Retrieval Augmented Generation](https://mlsanity.com/paper/2501.15915/) (2025)
- [Learn to Refuse: Making Large Language Models More Controllable and Reliable through Knowledge Scope Limitation and Refusal Mechanism](https://mlsanity.com/paper/2311.01041/) (2023)
- [Learning Conformal Abstention Policies for Adaptive Risk Management in Large Language and Vision-Language Models](https://mlsanity.com/paper/2502.06884/) (2025)
- [Supervising the search process produces reliable and generalizable information-seeking agents](https://mlsanity.com/paper/2502.13957/) (2025)
- [Empowering Working Memory for Large Language Model Agents](https://mlsanity.com/paper/2312.17259/) (2023)
- [Building Effective AI Coding Agents for the Terminal: Scaffolding, Harness, Context Engineering, and Lessons Learned](https://mlsanity.com/paper/2603.05344/) (2026)
- [Selective Generation for Controllable Language Models](https://mlsanity.com/paper/2307.09254/) (2023)
- [Optimal Query Allocation in Extractive QA with LLMs: A Learning-to-Defer Framework with Theoretical Guarantees](https://mlsanity.com/paper/2410.15761/) (2024)
- [From Code Foundation Models to Agents and Applications: A Comprehensive Survey and Practical Guide to Code Intelligence](https://mlsanity.com/paper/2511.18538/) (2025)
- [ContextBudget: Budget-Aware Context Management for Long-Horizon Search Agents](https://mlsanity.com/paper/2604.01664/) (2026)
- [A Contemporary Survey of Large Language Model Assisted Program Analysis](https://mlsanity.com/paper/2502.18474/) (2025)
- [Direct Retrieval-augmented Optimization: Synergizing Knowledge Selection and Language Models](https://mlsanity.com/paper/2505.03075/) (2025)
- [Repair Ingredients Are All You Need: Improving Large Language Model-Based Program Repair via Repair Ingredients Search](https://mlsanity.com/paper/2506.23100/) (2025)
- [Reliable Decision Making via Calibration Oriented Retrieval Augmented Generation](https://mlsanity.com/paper/2411.08891/) (2024)
- [Agentic Software Issue Resolution with Large Language Models: A Survey](https://mlsanity.com/paper/2512.22256/) (2025)
- [TreePS-RAG: Tree-based Process Supervision for Reinforcement Learning in Agentic RAG](https://mlsanity.com/paper/2601.06922/) (2026)
- [I-CALM: Incentivizing Confidence-Aware Abstention for LLM Selective Answering](https://mlsanity.com/paper/2604.03904/) (2026)
- [FLAIR: Feedback Learning for Adaptive Information Retrieval](https://mlsanity.com/paper/2508.13390/) (2025)
- [SIRAG: Towards Stable and Interpretable RAG with A Process-Supervised Multi-Agent Framework](https://mlsanity.com/paper/2509.18167/) (2025)

## Built on by (most cited first, 3 of 3)

- [Memory Reward Inflation in Self-Improving LLM Agents](https://mlsanity.com/paper/2608.00017/) (2026)
- [From Relevance to Execution Utility: Reward-Aware Dynamic Execution Gating for Skill-Based LLM Agents](https://mlsanity.com/paper/2608.09168/) (2026)
- [What Can Component-Replacement Evidence Establish? A Critical Scoping Review of Local Decisions in LLM Agents](https://mlsanity.com/paper/2609.39989/) (2026)

## Abstract (arXiv)

Large language model (LLM)-based coding agents increasingly rely on external memory to reuse prior debugging experience, repair traces, and repository-local operational knowledge. However, retrieved memory is useful only when the current failure is genuinely compatible with a previous one; superficial similarity in stack traces, terminal errors, paths, or configuration symptoms can lead to unsafe memory injection. This paper reframes issue-memory use as a selective, risk-sensitive control problem rather than a pure top-k retrieval problem. We introduce RSCB-MC, a risk-sensitive contextual bandit memory controller that decides whether an agent should use no memory, inject the top resolution, summarize multiple candidates, perform high-precision or high-recall retrieval, abstain, or ask for feedback. The system stores reusable issue knowledge through a pattern-variant-episode schema and converts retrieval evidence into a fixed 16-feature contextual state capturing relevance, uncertainty, structural compatibility, feedback history, false-positive risk, latency, and token cost. Its reward design penalizes false-positive memory injection more strongly than missed reuse, making non-injection and abstention first-class safety actions. In deterministic smoke-scale artifacts, RSCB-MC obtains the strongest non-oracle offline replay success rate, 62.5%, while maintaining a 0.0% false-positive rate. In a bounded 200-case hot-path validation, it reaches 60.5% proxy success with 0.0% false positives and a 331.466 microseconds p95 decision latency. The results show that, for coding-agent memory, the key question is not only which memory is most similar, but whether any retrieved memory is safe enough to influence the debugging trajectory.