Skip to content
NewA calmer, faster ZeroHour

Beyond Corrected Memory: Execution Consistency in Multi-Agent Systems

arXiv cs.AI / cs.LG / cs.CLby Zhe Yu ()1 min read4 · 2 reads
AI summary · grok-4.7Summaries by Vani AI

Researchers show correct shared memory can mask duty violations and introduce CAVERT for diagnosis.

The paper defines execution consistency for multi-agent systems through duties over state use, information handoffs, and final-state agreement, arguing that identical retained records can match both compliant and violating executions. Removing evidence such as receipts, action dependence, or response validity left 82.4% of opposite-label pairs indistinguishable, while restoration separated 97.9%. CAVERT extracts supported relationships from logs and outperformed contract-prompted LLM and rule-based baselines in all 12 benchmark-executor diagnosis settings, and beat rule-guided recovery in all four evaluated environments.

Key facts
  • Correct shared memory can correspond to both compliant and violating executions.
  • Removing key evidence left 82.4% of opposite pairs indistinguishable.
  • Restoring that evidence separated 97.9% of the merged pairs.
  • CAVERT led diagnosis across all 12 benchmark-executor settings.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

1 min read at arxiv.org

Shared memory coordinates agents' actions, but correct records do not establish that those actions satisfy task requirements. Memory governance and failure diagnosis regulate or inspect recorded information; they do not by themselves establish whether it is sufficient to judge task duties.

arXiv cs.AI / cs.LG / cs.CL

By Zhe Yu · arxiv.org

Read the full article at arXiv cs.AI / cs.LG / cs.CL

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.