Source record

Published by arXiv cs.MA (T1 source). The original is at https://arxiv.org/abs/2610.10126.

Pipeline notes

The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.

Paper presenting MAScope, a two-stage framework that recovers communication topology from multi-agent LLM execution traces and uses it to classify coordination failures. Experiments show topology context roughly doubles Macro-F1 for a small model and cuts diagnosis cost to about 6% of a high-quality baseline.

First systematic method to extract communication topology from LLM agent traces and use it as conditioning for failure diagnosis, with concrete accuracy and cost numbers.

Cited by

No citations on record.