Skip to content

feat(core): Add streamGenAiSpans options to stream gen_ai spans #20785

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
andreiborza merged 9 commits into from
May 12, 2026
Merged

feat(core): Add streamGenAiSpans options to stream gen_ai spans #20785

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
21aff86
feat(core): Add `streamGenAiSpans` options to stream gen_ai spans
andreiborza May 11, 2026
bbc19b7
Remove opting into gen ai span streaming for test apps that don't nee…
andreiborza May 11, 2026
e4d67bf
Merge array stringifying into the same loop as vercel.ai renaming
andreiborza May 11, 2026
17c96dc
Flip vercelai attribute manipulation order
andreiborza May 11, 2026
c770f57
Ensure we add ingest_settings when in browser
andreiborza May 11, 2026
623e6f1
Remove flag from test
andreiborza May 11, 2026
6bcd3ab
Add missing google operation name attribute
andreiborza May 11, 2026
cc2e598
Moved flag out of _experiments
andreiborza May 11, 2026
7fa8a76
Update changelog with a more explicit CTA
andreiborza May 12, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 17 additions & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,23 @@

Work in this release was contributed by @dmmulroy. Thank you for your contribution!

### Important Changes

- **feat(core): Add `streamGenAiSpans` options to stream gen_ai spans ([#20785](https://github.com/getsentry/sentry-javascript/pull/20785))**

Adds a new `streamGenAiSpans` option that controls how `gen_ai` spans are
sent to Sentry. When set, the SDK extracts all `gen_ai` spans out of a
transaction and sends them as v2 envelope items.
Copy link
Copy Markdown
Member

@JPeer264 JPeer264 May 11, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

q: Is it an important change if practically no one knows yet what a v2 envelope item is?

Copy link
Copy Markdown
Member Author

@andreiborza andreiborza May 11, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think so, the justification for it is in the next line:

This avoids payload size limits of regular transactions.


Enable this option if gen_ai spans are being dropped because the transaction payload exceeds size limits.

```ts
Sentry.init({
dsn: 'https://examplePublicKey@o0.ingest.sentry.io/0',
streamGenAiSpans: true,
});
```

## 10.52.0

### Important Changes
Expand Down
1 change: 1 addition & 0 deletions dev-packages/cloudflare-integration-tests/suites/tracing/anthropic-ai/index.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,7 @@ export default Sentry.withSentry(
(env: Env) => ({
dsn: env.SENTRY_DSN,
tracesSampleRate: 1.0,
streamGenAiSpans: true,
}),
{
async fetch(_request, _env, _ctx) {
Expand Down
51 changes: 28 additions & 23 deletions dev-packages/cloudflare-integration-tests/suites/tracing/anthropic-ai/test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,3 @@
import { SEMANTIC_ATTRIBUTE_SENTRY_OP, SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN } from '@sentry/core';
import { expect, it } from 'vitest';
import {
GEN_AI_OPERATION_NAME_ATTRIBUTE,
Expand All @@ -21,30 +20,36 @@ it('traces a basic message creation request', async ({ signal }) => {
const runner = createRunner(__dirname)
.ignore('event')
.expect(envelope => {
// Transaction item (first item in envelope)
const transactionEvent = envelope[1]?.[0]?.[1] as any;

expect(transactionEvent.transaction).toBe('GET /');
expect(transactionEvent.spans).toEqual(
expect.arrayContaining([
expect.objectContaining({
data: expect.objectContaining({
[GEN_AI_OPERATION_NAME_ATTRIBUTE]: 'chat',
[SEMANTIC_ATTRIBUTE_SENTRY_OP]: 'gen_ai.chat',
[SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN]: 'auto.ai.anthropic',
[GEN_AI_SYSTEM_ATTRIBUTE]: 'anthropic',
[GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'claude-3-haiku-20240307',
[GEN_AI_REQUEST_TEMPERATURE_ATTRIBUTE]: 0.7,
[GEN_AI_RESPONSE_MODEL_ATTRIBUTE]: 'claude-3-haiku-20240307',
[GEN_AI_RESPONSE_ID_ATTRIBUTE]: 'msg_mock123',
[GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 10,
[GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 15,
}),
description: 'chat claude-3-haiku-20240307',
op: 'gen_ai.chat',
origin: 'auto.ai.anthropic',
}),
]),
);

// Span container item (second item in same envelope)
const container = envelope[1]?.[1]?.[1] as any;
expect(container).toBeDefined();

expect(container.items).toHaveLength(1);
const [firstSpan] = container.items;

// [0] chat claude-3-haiku-20240307
expect(firstSpan!.name).toBe('chat claude-3-haiku-20240307');
expect(firstSpan!.status).toBe('ok');
expect(firstSpan!.attributes[GEN_AI_OPERATION_NAME_ATTRIBUTE]).toEqual({ type: 'string', value: 'chat' });
expect(firstSpan!.attributes['sentry.op']).toEqual({ type: 'string', value: 'gen_ai.chat' });
expect(firstSpan!.attributes['sentry.origin']).toEqual({ type: 'string', value: 'auto.ai.anthropic' });
expect(firstSpan!.attributes[GEN_AI_SYSTEM_ATTRIBUTE]).toEqual({ type: 'string', value: 'anthropic' });
expect(firstSpan!.attributes[GEN_AI_REQUEST_MODEL_ATTRIBUTE]).toEqual({
type: 'string',
value: 'claude-3-haiku-20240307',
});
expect(firstSpan!.attributes[GEN_AI_REQUEST_TEMPERATURE_ATTRIBUTE]).toEqual({ type: 'double', value: 0.7 });
expect(firstSpan!.attributes[GEN_AI_RESPONSE_MODEL_ATTRIBUTE]).toEqual({
type: 'string',
value: 'claude-3-haiku-20240307',
});
expect(firstSpan!.attributes[GEN_AI_RESPONSE_ID_ATTRIBUTE]).toEqual({ type: 'string', value: 'msg_mock123' });
expect(firstSpan!.attributes[GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 10 });
expect(firstSpan!.attributes[GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 15 });
})
.start(signal);
await runner.makeRequest('get', '/');
Expand Down
1 change: 1 addition & 0 deletions dev-packages/cloudflare-integration-tests/suites/tracing/google-genai/index.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,7 @@ export default Sentry.withSentry(
(env: Env) => ({
dsn: env.SENTRY_DSN,
tracesSampleRate: 1.0,
streamGenAiSpans: true,
}),
{
async fetch(_request, _env, _ctx) {
Expand Down
113 changes: 57 additions & 56 deletions dev-packages/cloudflare-integration-tests/suites/tracing/google-genai/test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,3 @@
import { SEMANTIC_ATTRIBUTE_SENTRY_OP, SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN } from '@sentry/core';
import { expect, it } from 'vitest';
import {
GEN_AI_OPERATION_NAME_ATTRIBUTE,
Expand All @@ -18,67 +17,69 @@ import { createRunner } from '../../../runner';
// want to test that the instrumentation does not break in our
// cloudflare SDK.

it('traces Google GenAI chat creation and message sending', async () => {
it('traces Google GenAI chat creation and message sending', async ({ signal }) => {
const runner = createRunner(__dirname)
.ignore('event')
.expect(envelope => {
// Transaction item (first item in envelope)
const transactionEvent = envelope[1]?.[0]?.[1] as any;

expect(transactionEvent.transaction).toBe('GET /');
expect(transactionEvent.spans).toEqual(
expect.arrayContaining([
// chat.sendMessage
expect.objectContaining({
data: expect.objectContaining({
[GEN_AI_OPERATION_NAME_ATTRIBUTE]: 'chat',
[SEMANTIC_ATTRIBUTE_SENTRY_OP]: 'gen_ai.chat',
[SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN]: 'auto.ai.google_genai',
[GEN_AI_SYSTEM_ATTRIBUTE]: 'google_genai',
[GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'gemini-1.5-pro',
[GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 8,
[GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 12,
[GEN_AI_USAGE_TOTAL_TOKENS_ATTRIBUTE]: 20,
}),
description: 'chat gemini-1.5-pro',
op: 'gen_ai.chat',
origin: 'auto.ai.google_genai',
}),
// models.generateContent
expect.objectContaining({
data: expect.objectContaining({
[GEN_AI_OPERATION_NAME_ATTRIBUTE]: 'generate_content',
[SEMANTIC_ATTRIBUTE_SENTRY_OP]: 'gen_ai.generate_content',
[SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN]: 'auto.ai.google_genai',
[GEN_AI_SYSTEM_ATTRIBUTE]: 'google_genai',
[GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'gemini-1.5-flash',
[GEN_AI_REQUEST_TEMPERATURE_ATTRIBUTE]: 0.7,
[GEN_AI_REQUEST_TOP_P_ATTRIBUTE]: 0.9,
[GEN_AI_REQUEST_MAX_TOKENS_ATTRIBUTE]: 100,
[GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 8,
[GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 12,
[GEN_AI_USAGE_TOTAL_TOKENS_ATTRIBUTE]: 20,
}),
description: 'generate_content gemini-1.5-flash',
op: 'gen_ai.generate_content',
origin: 'auto.ai.google_genai',
}),
// models.embedContent
expect.objectContaining({
data: expect.objectContaining({
[GEN_AI_OPERATION_NAME_ATTRIBUTE]: 'embeddings',
[SEMANTIC_ATTRIBUTE_SENTRY_OP]: 'gen_ai.embeddings',
[SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN]: 'auto.ai.google_genai',
[GEN_AI_SYSTEM_ATTRIBUTE]: 'google_genai',
[GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'text-embedding-004',
}),
description: 'embeddings text-embedding-004',
op: 'gen_ai.embeddings',
origin: 'auto.ai.google_genai',
}),
]),
);

// Span container item (second item in same envelope)
const container = envelope[1]?.[1]?.[1] as any;
expect(container).toBeDefined();
expect(container.items).toHaveLength(3);
const [firstSpan, secondSpan, thirdSpan] = container.items;

// [0] chat gemini-1.5-pro
expect(firstSpan!.name).toBe('chat gemini-1.5-pro');
expect(firstSpan!.status).toBe('ok');
expect(firstSpan!.attributes[GEN_AI_OPERATION_NAME_ATTRIBUTE]).toEqual({ type: 'string', value: 'chat' });
expect(firstSpan!.attributes['sentry.op']).toEqual({ type: 'string', value: 'gen_ai.chat' });
expect(firstSpan!.attributes['sentry.origin']).toEqual({ type: 'string', value: 'auto.ai.google_genai' });
expect(firstSpan!.attributes[GEN_AI_SYSTEM_ATTRIBUTE]).toEqual({ type: 'string', value: 'google_genai' });
expect(firstSpan!.attributes[GEN_AI_REQUEST_MODEL_ATTRIBUTE]).toEqual({
type: 'string',
value: 'gemini-1.5-pro',
});
expect(firstSpan!.attributes[GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 8 });
expect(firstSpan!.attributes[GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 12 });
expect(firstSpan!.attributes[GEN_AI_USAGE_TOTAL_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 20 });

// [1] generate_content gemini-1.5-flash
expect(secondSpan!.name).toBe('generate_content gemini-1.5-flash');
expect(secondSpan!.status).toBe('ok');
expect(secondSpan!.attributes[GEN_AI_OPERATION_NAME_ATTRIBUTE]).toEqual({
type: 'string',
value: 'generate_content',
});
expect(secondSpan!.attributes['sentry.op']).toEqual({ type: 'string', value: 'gen_ai.generate_content' });
expect(secondSpan!.attributes['sentry.origin']).toEqual({ type: 'string', value: 'auto.ai.google_genai' });
expect(secondSpan!.attributes[GEN_AI_SYSTEM_ATTRIBUTE]).toEqual({ type: 'string', value: 'google_genai' });
expect(secondSpan!.attributes[GEN_AI_REQUEST_MODEL_ATTRIBUTE]).toEqual({
type: 'string',
value: 'gemini-1.5-flash',
});
expect(secondSpan!.attributes[GEN_AI_REQUEST_TEMPERATURE_ATTRIBUTE]).toEqual({ type: 'double', value: 0.7 });
expect(secondSpan!.attributes[GEN_AI_REQUEST_TOP_P_ATTRIBUTE]).toEqual({ type: 'double', value: 0.9 });
expect(secondSpan!.attributes[GEN_AI_REQUEST_MAX_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 100 });
expect(secondSpan!.attributes[GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 8 });
expect(secondSpan!.attributes[GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 12 });
expect(secondSpan!.attributes[GEN_AI_USAGE_TOTAL_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 20 });

// [2] embeddings text-embedding-004
expect(thirdSpan!.name).toBe('embeddings text-embedding-004');
expect(thirdSpan!.status).toBe('ok');
expect(thirdSpan!.attributes[GEN_AI_OPERATION_NAME_ATTRIBUTE]).toEqual({ type: 'string', value: 'embeddings' });
expect(thirdSpan!.attributes['sentry.op']).toEqual({ type: 'string', value: 'gen_ai.embeddings' });
expect(thirdSpan!.attributes['sentry.origin']).toEqual({ type: 'string', value: 'auto.ai.google_genai' });
expect(thirdSpan!.attributes[GEN_AI_SYSTEM_ATTRIBUTE]).toEqual({ type: 'string', value: 'google_genai' });
expect(thirdSpan!.attributes[GEN_AI_REQUEST_MODEL_ATTRIBUTE]).toEqual({
type: 'string',
value: 'text-embedding-004',
});
})
.start();
.start(signal);
await runner.makeRequest('get', '/');
await runner.completed();
});
1 change: 1 addition & 0 deletions dev-packages/cloudflare-integration-tests/suites/tracing/langchain/index.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,7 @@ export default Sentry.withSentry(
(env: Env) => ({
dsn: env.SENTRY_DSN,
tracesSampleRate: 1.0,
streamGenAiSpans: true,
}),
{
async fetch(_request, _env, _ctx) {
Expand Down
84 changes: 38 additions & 46 deletions dev-packages/cloudflare-integration-tests/suites/tracing/langchain/test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,3 @@
import { SEMANTIC_ATTRIBUTE_SENTRY_OP, SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN } from '@sentry/core';
import { expect, it } from 'vitest';
import {
GEN_AI_OPERATION_NAME_ATTRIBUTE,
Expand All @@ -22,53 +21,46 @@ it('traces langchain chat model, chain, and tool invocations', async ({ signal }
const runner = createRunner(__dirname)
.ignore('event')
.expect(envelope => {
// Transaction item (first item in envelope)
const transactionEvent = envelope[1]?.[0]?.[1] as any;

expect(transactionEvent.transaction).toBe('GET /');
expect(transactionEvent.spans).toEqual(
expect.arrayContaining([
// Chat model span
expect.objectContaining({
data: expect.objectContaining({
[GEN_AI_OPERATION_NAME_ATTRIBUTE]: 'chat',
[SEMANTIC_ATTRIBUTE_SENTRY_OP]: 'gen_ai.chat',
[SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN]: 'auto.ai.langchain',
[GEN_AI_SYSTEM_ATTRIBUTE]: 'anthropic',
[GEN_AI_REQUEST_MODEL_ATTRIBUTE]: 'claude-3-5-sonnet-20241022',
[GEN_AI_REQUEST_TEMPERATURE_ATTRIBUTE]: 0.7,
[GEN_AI_REQUEST_MAX_TOKENS_ATTRIBUTE]: 100,
[GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]: 10,
[GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]: 15,
[GEN_AI_USAGE_TOTAL_TOKENS_ATTRIBUTE]: 25,
}),
description: 'chat claude-3-5-sonnet-20241022',
op: 'gen_ai.chat',
origin: 'auto.ai.langchain',
}),
// Chain span
expect.objectContaining({
data: expect.objectContaining({
[SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN]: 'auto.ai.langchain',
[SEMANTIC_ATTRIBUTE_SENTRY_OP]: 'gen_ai.invoke_agent',
'langchain.chain.name': 'my_test_chain',
}),
description: 'chain my_test_chain',
op: 'gen_ai.invoke_agent',
origin: 'auto.ai.langchain',
}),
// Tool span
expect.objectContaining({
data: expect.objectContaining({
[SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN]: 'auto.ai.langchain',
[SEMANTIC_ATTRIBUTE_SENTRY_OP]: 'gen_ai.execute_tool',
[GEN_AI_TOOL_NAME_ATTRIBUTE]: 'search_tool',
}),
description: 'execute_tool search_tool',
op: 'gen_ai.execute_tool',
origin: 'auto.ai.langchain',
}),
]),
);

// Span container item (second item in same envelope)
const container = envelope[1]?.[1]?.[1] as any;
expect(container).toBeDefined();
expect(container.items).toHaveLength(3);
const [firstSpan, secondSpan, thirdSpan] = container.items;

// [0] chat claude-3-5-sonnet-20241022
expect(firstSpan!.name).toBe('chat claude-3-5-sonnet-20241022');
expect(firstSpan!.status).toBe('ok');
expect(firstSpan!.attributes[GEN_AI_OPERATION_NAME_ATTRIBUTE]).toEqual({ type: 'string', value: 'chat' });
expect(firstSpan!.attributes['sentry.op']).toEqual({ type: 'string', value: 'gen_ai.chat' });
expect(firstSpan!.attributes['sentry.origin']).toEqual({ type: 'string', value: 'auto.ai.langchain' });
expect(firstSpan!.attributes[GEN_AI_SYSTEM_ATTRIBUTE]).toEqual({ type: 'string', value: 'anthropic' });
expect(firstSpan!.attributes[GEN_AI_REQUEST_MODEL_ATTRIBUTE]).toEqual({
type: 'string',
value: 'claude-3-5-sonnet-20241022',
});
expect(firstSpan!.attributes[GEN_AI_REQUEST_TEMPERATURE_ATTRIBUTE]).toEqual({ type: 'double', value: 0.7 });
expect(firstSpan!.attributes[GEN_AI_REQUEST_MAX_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 100 });
expect(firstSpan!.attributes[GEN_AI_USAGE_INPUT_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 10 });
expect(firstSpan!.attributes[GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 15 });
expect(firstSpan!.attributes[GEN_AI_USAGE_TOTAL_TOKENS_ATTRIBUTE]).toEqual({ type: 'integer', value: 25 });

// [1] chain my_test_chain
expect(secondSpan!.name).toBe('chain my_test_chain');
expect(secondSpan!.status).toBe('ok');
expect(secondSpan!.attributes['sentry.origin']).toEqual({ type: 'string', value: 'auto.ai.langchain' });
expect(secondSpan!.attributes['sentry.op']).toEqual({ type: 'string', value: 'gen_ai.invoke_agent' });
expect(secondSpan!.attributes['langchain.chain.name']).toEqual({ type: 'string', value: 'my_test_chain' });

// [2] execute_tool search_tool
expect(thirdSpan!.name).toBe('execute_tool search_tool');
expect(thirdSpan!.status).toBe('ok');
expect(thirdSpan!.attributes['sentry.origin']).toEqual({ type: 'string', value: 'auto.ai.langchain' });
expect(thirdSpan!.attributes['sentry.op']).toEqual({ type: 'string', value: 'gen_ai.execute_tool' });
expect(thirdSpan!.attributes[GEN_AI_TOOL_NAME_ATTRIBUTE]).toEqual({ type: 'string', value: 'search_tool' });
})
.start(signal);
await runner.makeRequest('get', '/');
Expand Down
1 change: 1 addition & 0 deletions dev-packages/cloudflare-integration-tests/suites/tracing/langgraph/index.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,7 @@ export default Sentry.withSentry(
dsn: env.SENTRY_DSN,
tracesSampleRate: 1.0,
sendDefaultPii: true,
streamGenAiSpans: true,
}),
{
async fetch(_request, _env, _ctx) {
Expand Down
Loading
Loading