Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Balakrishnan, Mahesh, Bharambe, Ashwin, Testuggine, Davide, Geraghty, David, Mao, David, Venkat, Vidhya, Mironov, Ilya, Baradi, Rithesh, Aiyer, Gayathri, Dudin, Victoria
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2604.07988
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866910115836198912
author Balakrishnan, Mahesh
Bharambe, Ashwin
Testuggine, Davide
Geraghty, David
Mao, David
Venkat, Vidhya
Mironov, Ilya
Baradi, Rithesh
Aiyer, Gayathri
Dudin, Victoria
author_facet Balakrishnan, Mahesh
Bharambe, Ashwin
Testuggine, Davide
Geraghty, David
Mao, David
Venkat, Vidhya
Mironov, Ilya
Baradi, Rithesh
Aiyer, Gayathri
Dudin, Victoria
contents Agents are LLM-driven components that can mutate environments in powerful, arbitrary ways. Extracting guarantees for the execution of agents in production environments can be challenging due to asynchrony and failures. In this paper, we propose a new abstraction called LogAct, where each agent is a deconstructed state machine playing a shared log. In LogAct, agentic actions are visible in the shared log before they are executed; can be stopped prior to execution by pluggable, decoupled voters; and recovered consistently in the case of agent or environment failure. LogAct enables agentic introspection, allowing the agent to analyze its own execution history using LLM inference, which in turn enables semantic variants of recovery, health check, and optimization. In our evaluation, LogAct agents recover efficiently and correctly from failures; debug their own performance; optimize token usage in swarms; and stop all unwanted actions for a target model on a representative benchmark with just a 3% drop in benign utility.
format Preprint
id arxiv_https___arxiv_org_abs_2604_07988
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle LogAct: Enabling Agentic Reliability via Shared Logs
Balakrishnan, Mahesh
Bharambe, Ashwin
Testuggine, Davide
Geraghty, David
Mao, David
Venkat, Vidhya
Mironov, Ilya
Baradi, Rithesh
Aiyer, Gayathri
Dudin, Victoria
Distributed, Parallel, and Cluster Computing
Artificial Intelligence
Agents are LLM-driven components that can mutate environments in powerful, arbitrary ways. Extracting guarantees for the execution of agents in production environments can be challenging due to asynchrony and failures. In this paper, we propose a new abstraction called LogAct, where each agent is a deconstructed state machine playing a shared log. In LogAct, agentic actions are visible in the shared log before they are executed; can be stopped prior to execution by pluggable, decoupled voters; and recovered consistently in the case of agent or environment failure. LogAct enables agentic introspection, allowing the agent to analyze its own execution history using LLM inference, which in turn enables semantic variants of recovery, health check, and optimization. In our evaluation, LogAct agents recover efficiently and correctly from failures; debug their own performance; optimize token usage in swarms; and stop all unwanted actions for a target model on a representative benchmark with just a 3% drop in benign utility.
title LogAct: Enabling Agentic Reliability via Shared Logs
topic Distributed, Parallel, and Cluster Computing
Artificial Intelligence
url https://arxiv.org/abs/2604.07988