feat(core): Enable streamGenAiSpans by default#21732
Conversation
Flip the `streamGenAiSpans` default to opt-out. The SDK now extracts gen_ai spans from transactions and sends them as v2 envelope items unless `streamGenAiSpans: false` is set. As before, streaming also disables gen_ai message truncation by default. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
d1684dd to
9d7aa86
Compare
nicohrubec
left a comment
There was a problem hiding this comment.
are we worried that this no completely shuts off truncation including image redaction? what we could do is already basically do the work to build two separate paths for image and text truncation and then with v11 we could cleanly rip out the text truncation part without having to do the separation then. then we could also use the image truncation for the streamGenAiSpans part (but we can also follow up on this)
|
|
||
| ### Important Changes | ||
|
|
||
| - **feat(core): Enable `streamGenAiSpans` by default ([#21732](https://github.com/getsentry/sentry-javascript/pull/21732))** |
There was a problem hiding this comment.
should we even call this out? or maybe shorten this to basically only call out the bit about truncation? because that's the only actionable part for users
There was a problem hiding this comment.
I think it's also actionable to know what this does so they can make a decision to turn it off instead of having to go find what this flag even is.
|
No we are not worried about media truncation being off by default. We will document ways to redact media in |
Flip the
streamGenAiSpansdefault to opt-out. The SDK now extracts gen_ai spans from transactions and sends them as v2 envelope items unlessstreamGenAiSpans: falseis set. As when the option was explicitly enabled, this disables gen_ai message truncation by default; setenableTruncation: trueon the respective AI integration to opt back in.Self-hosted Sentry users should set
streamGenAiSpans: false, since streamed gen_ai spans may not be ingested by their instance.