Mnemosyne: Agentic Transaction Processing for Validating and Repairing AI-generated Workflows
Quick Answer
Mnemosyne introduces Agentic Transaction Processing (ATP) to validate AI-generated workflows, ensuring actions are trustworthy before execution.
Quick Take
It features a runtime with an append-only log and achieves under 6% overhead in projection and validation, while local repairs require significantly fewer operations than global recompute.
Key Points
- ATP treats generated actions as untrusted proposals until they meet executable constraints.
- Mnemosyne proves four safety properties, including authority separation and evidence-preserving repair.
- Local repair edits require an order of magnitude fewer operations than global recompute.
- The system maintains a bounded-reactive-repair guarantee for localized repair protocols.
- Mnemosyne is open source, available on GitHub.
Paper Resources
📖 Reader Mode
~2 min readAbstract:LLMs, solvers, and agent teams increasingly generate workflow actions, repairs, and plans, but a generated action may be syntactically valid yet stale, infeasible, conflicting, or destructive of the evidence that triggered a repair. We introduce Agentic Transaction Processing (ATP), a transaction model that treats generated actions as untrusted proposals until they pass deterministic admission under a declared, executable constraint set C. The principle is two-sided: a proposal is not truth, and no proposal foresees every disruption: anything may propose, but only the runtime admits and commits, and when an unforeseen disruption strikes it repairs reactively within bounds rather than trusting a fresh proposal. Relative to C, committed-state correctness becomes independent of the competence, honesty, or learning of the proposing layer. We realize ATP in Mnemosyne, a runtime with an append-only transition log, effective-state projection, dependency-safe compensation, and active commitment records, and prove four safety properties relative to C (authority separation, serial-equivalent generative admission, evidence-preserving repair, and obligation containment) together with a bounded-reactive-repair guarantee for its localized repair protocol (LCRP). A reproducible artifact rejects the targeted violations across nine falsification tests while still admitting valid work, at under 6% projection-and-validation overhead, and bounded local repair edits an order of magnitude fewer operations than global recompute. Mnemosyne is open source: this https URL.
| Comments: | 36 pages, 24 tables, 6 figures |
| Subjects: | Artificial Intelligence (cs.AI) |
| ACM classes: | H.2.4; I.2.7; C.2.4 |
| Cite as: | arXiv:2607.00269 [cs.AI] |
| (or arXiv:2607.00269v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.00269 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Edward Chang [view email]
[v1]
Tue, 30 Jun 2026 23:33:16 UTC (56 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →HOBA: Hierarchical On-Policy Bidding Agents for Adaptive Online Advertising
HOBA (Hierarchical On-policy Bidding Agents) is a novel hierarchical reinforcement learning framework that enhances online advertising bidding systems by improving adaptability and reducing hyperparameter tuning costs. It utilizes a for hyperparameter inference, a SARSA agent for expert model selection, and a dynamic expert pool for bid execution, achieving a +3.6% increase in target cost during large-scale deployment and outperforming state-of-the-art baselines on AuctionNet.