Shared Selective Persistent Memory for Agentic LLM Systems
On the map: LLM agents and tool use → Agent memory systems
Abstract (arXiv)
Agentic LLM systems that generate code through multi-turn tool use face a fundamental context problem: each session starts from zero, discarding the domain constraints, data schemas, tool configurations, and output preferences that made previous sessions productive. We introduce shared selective persistent memory, an architecture that retains four categories of reusable context - task specifications, data schemas, tool configurations, and output constraints - while discarding session-specific reasoning traces, and that packages them into workspaces transferable across users under role-based access control. The resulting cost curve is non-monotonic. In a controlled replication on four public datasets, where a formatting specification is established once and then withheld, no memory completes 0/12 trials at 3.8K input tokens, selective memory completes 12/12 at 3.9K, and full conversation history completes 8/12 at 7.7K. What is kept matters more than how much is kept: the winning configuration costs essentially what the failing one does, and twice as much context does not improve on it. Both differences from no memory survive Bonferroni-corrected exact McNemar tests (p = 0.0005, p = 0.008); the two memory conditions separate on price rather than completion. We implement this in a deployed platform where agents produce git-versioned artifacts from CSV, SQL, REST, and MCP sources. A complementary zero-token data refresh contract decouples generated programs from runtime data, firing on 12/12 trials at a median 0.08s with no model call, while summary-driven data representation costs 97-431x fewer tokens than raw injection. Across 24 recurring enterprise tasks selective memory completes 23/24 against 19/24 and 17/24, though at that sample no pairwise difference reaches significance.
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 12 of its 24 references that are in the map. First posted on arXiv 10 Jul 2026.
Builds on
12 of 24 references.
Closest ancestors · 2
Same direction, newest first
- 2025ACON: Optimizing Context Compression for Long-horizon LLM Agents1 Oct
- 2025A-MEM: Agentic Memory for LLM Agents17 Feb
Other references · 10
- 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,160
- 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,355
- 2023Toolformer: Language Models Can Teach Themselves to Use Toolscited by 4,072
- 2023Generative Agents: Interactive Simulacra of Human Behaviorcited by 3,500
- 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,383
- 2023Gorilla: Large Language Model Connected with Massive APIscited by 1,302
- 2023MemGPT: Towards LLMs as Operating Systemscited by 1,295
- 2023Extending Context Window of Large Language Models via Positional Interpolationcited by 721
- 2023Language Agent Tree Search Unifies Reasoning Acting and Planning in Language Modelscited by 551
- 2025Reducing Cost of LLM Agents with Trajectory Reductioncited by 44
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026Memory as Infrastructure: Reliability Engineering for Persistent Agent Memory in Months-Long LLM-Assisted Development
- 2025Collaborative Memory: Multi-User Memory Sharing in LLM Agents with Dynamic Access Control
- 2026Agent Memory: Characterization and System Implications of Stateful Long-Horizon Workloads
- 2026Akashic: A Low-Overhead LLM Inference Service with MemAttention
- 2026Are We Ready For An Agent-Native Memory System?
- 2025MemTool: Optimizing Short-Term Memory Management for Dynamic Tool Calling in LLM Agent Multi-Turn Conversations
- 2026MemGym: a Long-Horizon Memory Environment for LLM Agents
- 2026Governed Collaborative Memory as Artificial Selection in LLM-Based Multi-Agent Systems