Rohan Paul (@rohanpaul_ai) · 2026/10/13 00:22:03
DigitalOcean 推出 Agent 静默故障检测：用规则捕获异常会话并自动回放验证修复
DigitalOcean 发布新工具组合，旨在解决 AI 智能体“静默失败”难题。通过轻量级规则引擎（Signals）实时监测用户重复提问、人工介入请求等异常信号，并将这些真实故障会话自动转化为测试集，利用模拟环境（Simulations）在部署前验证修复效果，替代了昂贵且低效的人工日志审查或全量 LLM 评分方案。
报道全文原始报道全文
Rohan Paul (@rohanpaul_ai) · 2026-10-12 16
Agent failures often pass every health check. A tool call hits the wrong database region, returns 200, and the user gets a slower, worse answer. Nothing alerts.
DigitalOcean’s new tools flag agent sessions that fail silently, then replay them against a fix before release.
Because, the clue behind an agent bug is often sitting in something nobody can search.
Signals watches every live session with cheap rules, not an AI model, that look for human trouble signs such as a user repeating themselves, asking for a human, or a tool being retried in a loop, then passes those sessions to Simulations, which replays them with a fake user against your fixed agent to check the whole conversation works before the fix ships.
Until now teams had to pay an LLM to grade every conversation, read transcripts by hand, or invent test cases from imagination, while here the real failures from production automatically become the test set.
↳ 引用 @digitalocean (DigitalOcean): Catch it at the first user, not the four-hundredth.
A new user’s bank link fails, the agent says “try again later,” and nothing shows up in the logs. Watch a team find the bug and fix it before anyone else hits it, with Agent Learning Services and Agent Data Services, launching today.
