# ActionRAG: Evidence-Refining Graph Diffusion for Long-Term Memory in LLM Agents

DevFeed: [ActionRAG: Evidence-Refining Graph Diffusion for Long-Term Memory in LLM Agents](<https://devfeed.tech/articles/actionrag-evidence-refining-graph-diffusion-for-long-term-memory-in-llm-agents-100042.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.32752>)

Author: Navyansh Mahla, Sashank Vandrangi

Published: 2026-10-26T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Graphs](<https://devfeed.tech/topics/graphs.md>), [Entity resolution](<https://devfeed.tech/topics/entity-resolution.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [diffusion](<https://devfeed.tech/tags/diffusion.md>), [entity-resolution](<https://devfeed.tech/tags/entity-resolution.md>), [flux](<https://devfeed.tech/tags/flux.md>), [graph](<https://devfeed.tech/tags/graph.md>), [indexing](<https://devfeed.tech/tags/indexing.md>), [learning](<https://devfeed.tech/tags/learning.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llm-agents](<https://devfeed.tech/tags/llm-agents.md>), [materialization](<https://devfeed.tech/tags/materialization.md>), [memory](<https://devfeed.tech/tags/memory.md>), [memory-architecture](<https://devfeed.tech/tags/memory-architecture.md>), [merge](<https://devfeed.tech/tags/merge.md>), [online](<https://devfeed.tech/tags/online.md>), [pipeline](<https://devfeed.tech/tags/pipeline.md>), [pipelines](<https://devfeed.tech/tags/pipelines.md>), [points](<https://devfeed.tech/tags/points.md>), [protection](<https://devfeed.tech/tags/protection.md>), [query](<https://devfeed.tech/tags/query.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>)

## AI overview

ActionRAG combines cautious graph indexing with query-time retrieval refinement to recover consistent evidence from an LLM agent's growing interaction history. It reconciles type labels and uses tiered entity resolution, then retrieves entities and triples to build a bounded local graph. Query-conditioned diffusion adjusts triple scores using local paths, while seed protection and hub-penalized flux preserve direct evidence and help recover connections. On MemoryAgentBench, it reports 55.1 average accuracy across eight deterministic-scoring columns, 5.4 points above the strongest retrieval-based memory baseline. Mean synthesis cost is below $0.001 per query on four reported cost columns; separately measured retrieval-side p99 components sum to 440.7 ms on a consumer Apple Silicon setup.

## Source excerpt

arXiv:2610.32752v1 Announce Type: new Abstract: LLM agents need long-term memory that can recover precise and mutually consistent evidence from growing interaction histories, particularly when answers depend on relations among facts rather than isolated semantic matches. Current memory pipelines remain vulnerable at two stages: indexing errors can fragment repeated entities or merge distinct ones, and online retrieval often treats initial similarity scores as final even when the surrounding graph provides evidence for revising them. We introduce ActionRAG, a graph memory architecture that couples cautious indexing with query-time graph retroevaluation. Its indexing pipeline reconciles type labels and applies tiered entity resolution. Given a query, dual-channel entity and triple retrieval forms an initial evidence set and induces a bounded local graph. Query-conditioned receiver-normalized diffusion refines triple scores using local path structure. Seed protection, hub-penalized flux, and triple materialization preserve direct evidence while recovering missed connections. We evaluate ActionRAG on MemoryAgentBench (MAB), which tests accurate retrieval, conflict resolution, long-range understanding, and test-time learning. Across its eight deterministic-scoring columns, ActionRAG achieves 55.1 average accuracy, improving over the strongest retrieval-based memory baseline by 5.4 points. Under MAB's query-time cost accounting, mean synthesis cost stays below \$0.001 per query on the four reported cost columns, and the sum of separately measured retrieval-side p99 components is 440.7 ms on our consumer Apple Silicon setup.