Tag
OpenCall 是一套 on-call agent 工作流剧本:PagerDuty 等告警触发后,Agent Swarm 在 Slack 开线程调查代码与遥测,经人工一键批准后自动开 PR 修复,并在 Claude、Codex、OpenCode 等 harness 间自动故障切换。
Introduces ORCA-bench, a production-fidelity benchmark for evaluating LLM agents on oncall root cause analysis, finding that even frontier agents achieve only 25.3% accuracy on medium-difficulty tasks.