Causal Episodic Memory for Feedback-Driven Agent Repair

arXiv 2608.05906 ↗First posted 6 Aug 2026cs.CLcited by 0

On the map: LLM agents and tool use → Agent memory systems

Abstract (arXiv)

LLM agents that repair failures often discard successful corrections, forcing later episodes to rediscover similar solutions. We study whether finalized repair outcomes can improve subsequent Text-to-SQL episodes without parameter updates. We introduce MERIT, a training-free agent that maintains an online dual-polarity memory of oracle-verified corrections and observed unsuccessful directions. Under oracle-assisted benchmark feedback, only memories from earlier finalized episodes are eligible for retrieval. A deterministic classifier assigns a coarse failure type, which conditions a hybrid lexical-dense retriever before the frozen model generates each revision. Using Qwen2.5-7B-Instruct with identical initial predictions and repair budgets, method improves execution accuracy over stateless iterative repair from (66.34%) to (69.79%) on Spider and from (47.35%) to (48.44%) on BIRD. Paired analyses provide clear evidence for the Spider gain but weaker evidence on BIRD. MERIT is not reliably separated from untyped dynamic retrieval on either benchmark, while Reflexion-style memory reaches (51.24%) on BIRD at substantially higher inference cost. Ablations show that negative memory contributes modestly, the value of type conditioning and lexical-dense ranking is dataset dependent, and schema-local experience provides the most consistent benefit. These results clarify when causal cross-query memory improves repair and when broader memory representations remain preferable. Our implementation is available here:

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Agent memory systems, one of 16 directions in LLM agents and tool use. Builds on 18 of its 27 references that are in the map. First posted on arXiv 6 Aug 2026.

Builds on

18 of 27 references.

References · 18

2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,355
2023Generative Agents: Interactive Simulacra of Human Behaviorcited by 3,500
2023Self-Refine: Iterative Refinement with Self-Feedbackcited by 3,317
2023MemGPT: Towards LLMs as Operating Systemscited by 1,295
2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 965
2023Teaching Large Language Models to Self-Debugcited by 963
2023ExpeL: LLM Agents Are Experiential Learnerscited by 814
2023CRITIC: Large Language Models Can Self-Correct with Tool-Interactive Critiquingcited by 751
2023DIN-SQL: Decomposed In-Context Learning of Text-to-SQL with Self-Correctioncited by 465
2023Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluationcited by 378
2024CHASE-SQL: Multi-Path Reasoning and Preference Optimized Candidate Selection in Text-to-SQLcited by 132
2023ACT-SQL: In-Context Learning for Text-to-SQL with Automatically-Generated Chain-of-Thoughtcited by 65

6 more

2024MAGIC: Generating Self-Correction Guideline for In-Context Text-to-SQLcited by 34
2025SQL-o1: A Self-Reward Heuristic Dynamic Search Method for Text-to-SQLcited by 19
2025Understanding, Detecting, and Repairing Real-World In-Context-Learning-Based Text-to-SQL Errorscited by 14
2025SQLens: An End-to-End Framework for Error Detection and Correction in Text-to-SQLcited by 10
2025ReEx-SQL: Reasoning with Execution-Aware Reinforcement Learning for Text-to-SQLcited by 8
2026Memo-SQL: Structured Decomposition and Experience-Driven Self-Correction for Training-Free NL2SQLcited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way