OpenRCA and rca-llm
OpenRCA targets root cause analysis in software failures using LLMs as the analytical tool, while rca-llm provides an evaluation framework specifically for assessing RCA performance in LLM inference systems themselves—making them **complements** that address different layers (RCA for general software vs. evaluation of RCA in LLM deployments).
About OpenRCA
microsoft/OpenRCA
[ICLR'25] OpenRCA: Can Large Language Models Locate the Root Cause of Software Failures?
Comprises a benchmark dataset across three production systems (Telecom, Bank, Market) with multi-modal telemetry including KPI time series, dependency traces, and semi-structured logs. Introduces RCA-agent, a Python-based agentic baseline that retrieves and analyzes telemetry programmatically rather than processing full contexts, reducing token overhead while maintaining reasoning capability. Supports custom task generation and evaluation with standardized metrics for assessing LLM performance on root cause localization in complex distributed systems.
About rca-llm
exalsius/rca-llm
An evaluation framework for root cause analysis in large-scale LLM inference systems
Scores updated daily from GitHub, PyPI, and npm data. How scores work