Features
Agent failure detection via Slack integration
Automated trace analysis and root cause identification
Agent swarm triage of production issues
Agent Judge for long-horizon agent evaluation
Compare runs to validate fixes before shipping
Production data replay for testing
Multi-session failure pattern discovery
Automated behavior tracking and recurrence alerts
Slack-native investigation interface
Custom eval criteria creation
Integration with LangSmith (inferred from docs)
Data Agent for data analysis with Python/SQL execution
Plugins Agent integrating 200+ third-party plugins
Web Agent for autonomous web browsing via Chrome extension
Auto plugin selection by the agent
Combined plugin usage for multi-step tasks
Chat-based web UI optimized for swift responses
Open-source full-stack code (backend + frontend)
Local deployment via Docker
Free online demo on OpenAgents website
Data visualization using Python libraries
Integration with 200+ plugins including Klarna, XWeather, Wolfram Alpha
Lemur foundation models for language agents (open-source)
Real-world evaluation and research support
Troubleshooting support via Discord
COLM 2024 publication with paper available
