Back to directory
Agents_Failure_Attribution
Free365Benchmark for automated failure attributions in agentic systems (🏆 ICML 2025 Spotlight)
About
This tool identifies why AI agents fail by analyzing their actions and outcomes. It's built for researchers testing autonomous systems. The benchmark includes real failure cases that let you measure how well different attribution methods pinpoint the root causes of agent errors.
Key Features
- agent
- autogen
- llm
- multi-agent
- multiagent-systems
Pricing
FreeOpen source. You supply your own LLM API keys.
Categories
Research