Can the Convex Sentry integration automatically integrate Sentry Monitor AI Agents?

[ethan-huo]

Convex integration is very easy to configure, but I guess it won’t automatically integrate https://a1dai.sentry.io/insights/ai/agents/?integration=vercelai&project=4509842103795712&statsPeriod=7d? Could you provide an automatic configuration feature for Monitor AI Agents? I urgently need to see complete telemetry data from LLM requests—an automatic integration would be fantastic!

[ianmacartney]

Can you extract what you need from the raw request/response handler? It has the request start/stop time, as well as what agent ran it. If I made it generic on the CustomCtx, then you could pass through any fields from the Agent call's `ctx` to get more fine-grain scoping? It looks like a lot of the metadata they have in their integration is already available if we could submit it - agent name, etc.

…

On Tue, Aug 19, 2025 at 6:07 AM Ethan Huo ***@***.***> wrote: *ethan-huo* created an issue (get-convex/agent#124) <#124> Convex integration is very easy to configure, but I guess it won’t automatically integrate https://a1dai.sentry.io/insights/ai/agents/?integration=vercelai&project=4509842103795712&statsPeriod=7d? Could you provide an automatic configuration feature for Monitor AI Agents? I urgently need to see complete telemetry data from LLM requests—an automatic integration would be fantastic! — Reply to this email directly, view it on GitHub <#124>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AACZQW73KKDZTYOMYQTVZJT3OMOSDAVCNFSM6AAAAACEIHNHJKVHI2DSMVQWIX3LMV43ASLTON2WKOZTGMZTIMZSGI2DCMI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[ethan-huo]

I'm not sure what the lifecycle of rawRequestResponseHandler is, as I haven't read the relevant code. Sentry's integration with the AI SDK should collect data through OTEL, and the AI SDK will emit a complete OTEL trace for the streaming generation process. I think for Convex, could we first add support for Sentry.vercelAIIntegration in the Sentry integration section of the project settings? As for other context, we can attach it within streamAsync.

Sentry.init({
  dsn: "...",
  integrations: [
    // Add the Vercel AI SDK integration 
    Sentry.vercelAIIntegration({
      recordInputs: true,
      recordOutputs: true,
    }),
  ],
  // Tracing must be enabled for agent monitoring to work
  tracesSampleRate: 1.0,
  sendDefaultPii: true,
});

    const result = await thread.streamText(
      {
        promptMessageId,
        experimental_telemetry: {
          isEnabled: true,
          recordInputs: true,
          recordOutputs: true,
          metadata: {agentName, threadId,  ...etc },
        },