fix(node): Use context-scoped suppression for LangChain + AI provider span deduplication#22366
Open
nicohrubec wants to merge 2 commits into
Open
fix(node): Use context-scoped suppression for LangChain + AI provider span deduplication#22366nicohrubec wants to merge 2 commits into
nicohrubec wants to merge 2 commits into
Conversation
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
Contributor
size-limit report 📦
|
nicohrubec
marked this pull request as ready for review
July 17, 2026 14:21
nicohrubec
requested review from
andreiborza and
isaacs
and removed request for
a team
July 17, 2026 14:21
nicohrubec
marked this pull request as draft
July 17, 2026 15:24
isaacs
approved these changes
Jul 17, 2026
isaacs
left a comment
isaacs
left a comment
Member
There was a problem hiding this comment.
LGTM! Some tiny nits / suggestions for a doc comment, but an improvement for sure.
| /** | ||
| * Execute a callback with AI provider spans suppressed in the current scope. | ||
| * This is used by higher-level integrations (like LangChain) to prevent | ||
| * duplicate spans from underlying AI provider instrumentations. |
Member
There was a problem hiding this comment.
low: I think this approach means that anything that calls getCurrentScope() will get this child scope instead of the original? It seems like it might have side effects if anything in the callback calls Sentry.setTag/setContext/setUser, since it'll be attached to a discarded scope.
I don't really see a way around it, tbh, but maybe a doc comment could be worthwhile here?
Suggested change
| * duplicate spans from underlying AI provider instrumentations. | |
| * duplicate spans from underlying AI provider instrumentations. | |
| * | |
| * Suppression rides on a forked scope (as `suppressTracing` does) because a | |
| * forked scope is the only Sentry-native carrier that both survives async | |
| * boundaries and stays isolated from concurrent, unrelated work. | |
| * | |
| * Trade-off: current-scope mutations made inside `callback` land on the fork | |
| * and do not propagate to the outer scope. |
| // EDGE CASE: Import and instantiate Anthropic client BEFORE LangChain is imported | ||
| // This simulates the timing issue where a user creates an Anthropic client in one file | ||
| // before importing LangChain in another file | ||
| // Direct Anthropic call made BEFORE LangChain is imported/used |
Member
There was a problem hiding this comment.
nit: this file is named scenario-openai-before-langchain.mjs but it is using anthropic? Should it be renamed scenario-anthropic-before-langchain.mjs? It doesn't change it functionally, of course, but just seems a little weird 😅
| // EDGE CASE: Import and instantiate Anthropic client BEFORE LangChain is imported | ||
| // This simulates the timing issue where a user creates an Anthropic client in one file | ||
| // before importing LangChain in another file | ||
| // Direct Anthropic call made BEFORE LangChain is imported/used |
| } | ||
|
|
||
| /** @inheritDoc */ | ||
| protected _setupIntegrations(): void { |
Member
There was a problem hiding this comment.
Nice. Removing this is a good cleanup.
nicohrubec
marked this pull request as ready for review
July 21, 2026 08:17
Member
Author
|
I noticed that this solution doesn't work as is for the orchestrion path. I will fix that in a follow up but need to put some of the global skip logic for that one back in for now. |
nicohrubec
force-pushed
the
feat/langchain-scoped-ai-suppression
branch
from
July 21, 2026 08:41
f522c23 to
099efb6
Compare
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 1 potential issue.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Want reviews to match your repository better? Bugbot Learning can learn team-specific rules from PR activity. A team admin can enable Learning in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit 099efb6. Configure here.
… span deduplication Replaces the global AI-provider skip flag with context-scoped suppression for the OTel LangChain path: LangChain invoke/stream/batch (and embeddings) run inside `_INTERNAL_withSuppressedAiProviderSpans`, and the core provider instrumentations skip span creation when `_INTERNAL_isAiProviderSpanSuppressed()`. Direct provider SDK calls outside LangChain keep their spans, fixing globally-dropped provider spans. The orchestrion provider channels continue to use the global skip flag for now; a follow-up converts them to scoped suppression. Fixes #19687 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
nicohrubec
force-pushed
the
feat/langchain-scoped-ai-suppression
branch
from
July 21, 2026 08:56
099efb6 to
7eed185
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
When both LangChain and an AI provider integration (OpenAI, Anthropic, Google GenAI) are active, a single LangChain invoke() call produces duplicate gen_ai.chat spans — one from LangChain's callback handler and one from the underlying provider instrumentation.
The previous fix globally disabled provider instrumentations when LangChain was detected, but this broke direct provider SDK calls unrelated to LangChain.
This PR sets a suppression flag on the current scope during LangChain invoke/stream/batch execution. Provider instrumentations check this flag and skip span creation if set.
Closes #19687