arXiv cs.CL· Yiqi Wang, Jinqian Ju, Jiaqi Zhang, Zhangkai Wu, Yiqun Duan, Zequn Sun, Taotao Cai, Mingkai Zheng·· 13 小时前AI 评分38
Tether：为工具调用型 LLM 智能体提供依赖绑定的记忆机制
Memories That Outlive Their Reasons: Dependency-Tethered Memory for Tool-Using LLM Agents
AI 导读
针对工具调用型 LLM 智能体的记忆失效问题，研究者提出 Tether——写入时记录每条结论所依赖的确切版本，读取时撤回任何支撑已失效的结论。在 Fallout 模拟环境中，Tether 的上游失效对象发布率仅 0.3%，验证调用比全量检查少 37%；而 Mem0 发布率达 42%，A-Mem 与 Graphiti 为 9%。
正文 · 原文
Abstract:Agents now act on the Web, publishing packages, models and datasets through registry APIs, and remembering past verifications spares them the work. But a verdict rests on premises. When a signing key is revoked, everything built on it stops passing, yet the advisory names only the key, and the old verdicts of its dependents may never share a retrieval window with it. This non-locality of invalidation turns memory into a liability: in long simulated chats, an agent reading a memory of atomic facts publishes 62% of the objects whose failure lies upstream, each backed by a record that was true when written. Yet derivation is local even when invalidation is not. Tether records, at write time, the exact versions each verdict derives from, and at read time withdraws any verdict whose support has failed anywhere in the store; the agent reads the same fields as before. On Fallout, a generator of versioned worlds with hypothesis-blind simulated histories shared by every memory, Tether publishes 0.3% of upstream-failing objects across three domains when relations are stated with their verdicts, with 37% fewer verification calls than checking everything. A-Mem and Graphiti re-verify most candidates (0.83--0.86 checks per candidate), which limits their savings, yet publish 9% of upstream-failing objects; Mem0 relies on memory more and publishes 42%. Flushing every verdict on any failure is as safe as Tether but costs 32% more checks. The gain over fact memory holds with three open-weight backbones, a whole-store window and real Go advisories. These gains depend on capturing dependencies: when relations arrive in sessions apart from their verdicts, Tether's downstream unsafe publication rate rises to 9.4--22.1%. It also degrades with weaker writers that miss dependencies or invalidation events.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.25123 [cs.CL] |
| (or arXiv:2610.25123v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.25123 arXiv-issued DOI via DataCite |
Submission history
From: Yiqi Wang [view email]
[v1]
Sun, 18 Oct 2026 09:01:52 UTC (313 KB)
[v2]
Thu, 22 Oct 2026 04:32:22 UTC (371 KB)
来源：arXiv cs.CL · arxiv.org