fix(sdk): proper formatting for vercel AI SDK tool calls #736
Conversation
|
Caution Review failedThe pull request is closed. Note Other AI code review bot(s) detectedCodeRabbit has detected other AI code review bot(s) in this pull request and will avoid duplicating their findings in the review comments. This may lead to a less comprehensive review. WalkthroughSplit AI SDK span processing into separate name and attribute transforms. Span names are remapped (including a TOOL_SPAN_NAME for tool calls), telemetry metadata is extracted into association properties, name transforms run on span start and attribute transforms run on span end. Public APIs and types were reorganized. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
actor App
participant SpanProcessor
participant AiTransforms as ai-sdk-transformations
participant Exporter
App->>SpanProcessor: onSpanStart(span)
SpanProcessor->>AiTransforms: transformAiSdkSpanNames(span)
Note right of AiTransforms #eef2ff: Map ai.* names → canonical names\nMark tool-call spans (TOOL_SPAN_NAME)
App->>SpanProcessor: onSpanEnd(span)
SpanProcessor->>AiTransforms: transformAiSdkSpanAttributes(span)
Note right of AiTransforms #fff7e6: Extract ai.telemetry.metadata.* → association props\nApply LLM attribute transforms and tool args/results mapping
SpanProcessor->>Exporter: proceed with export / OTLP compatibility
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Possibly related PRs
Suggested reviewers
Poem
Pre-merge checks and finishing touches✅ Passed checks (3 passed)
📜 Recent review detailsConfiguration used: CodeRabbit UI Review profile: CHILL Plan: Pro Disabled knowledge base sources:
📒 Files selected for processing (4)
Comment |
![ellipsis-dev[bot]](https://avatars.githubusercontent.com/in/64358?s=60&v=4){kind=small}
There was a problem hiding this comment.
Caution
Changes requested ❌
Reviewed everything up to 65c57a8 in 2 minutes and 27 seconds. Click for details.
- Reviewed
142lines of code in2files - Skipped
0files when reviewing. - Skipped posting
0draft comments. View those below. - Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.
Workflow ID: wflow_Xq8XoJByBky9HnEq
You can customize 
|
|
||
| export const transformAiSdkSpanNames = (span: Span): void => { | ||
| if (span.name === TOOL_SPAN_NAME) { | ||
| span.updateName(`${span.attributes["ai.toolCall.name"] as string}.tool`); |
There was a problem hiding this comment.
Guard the use of 'ai.toolCall.name' in transformAiSdkSpanNames to avoid formatting to 'undefined.tool' if the attribute is missing.
| span.updateName(`${span.attributes["ai.toolCall.name"] as string}.tool`); | |
| span.updateName(`${span.attributes["ai.toolCall.name"] ? span.attributes["ai.toolCall.name"] : "tool"}.tool`); |
|
|
||
| const shouldHandleSpan = (span: ReadableSpan): boolean => { | ||
| return span.name in HANDLED_SPAN_NAMES; | ||
| return span.instrumentationScope.name === "ai"; |
There was a problem hiding this comment.
Use optional chaining for instrumentationScope (e.g. span.instrumentationScope?.name) in shouldHandleSpan to avoid potential runtime errors.
| return span.instrumentationScope.name === "ai"; | |
| return span.instrumentationScope?.name === "ai"; |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
packages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts (1)
366-376: Stringify non-primitive values before assigning to SpanAttributes
- packages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts — transformResponseText (lines 56–57) and transformResponseObject (lines 79–80): currently assign attributes[
${SpanAttributes.LLM_COMPLETIONS}.0.content] = attributes[AI_RESPONSE_*]; ensure value is a string (e.g. typeof v === 'string' ? v : JSON.stringify(v)) before assigning.- transformResponseToolCalls (function ~lines 100–137): attributes[
${SpanAttributes.LLM_COMPLETIONS}.0.tool_calls.${index}.arguments] = toolCall.args and the tool_call.arguments pushed into toolCallParts — stringify toolCall.args (or conditionally stringify) before assigning/pushing.- transformPrompts (line 263): attributes[contentKey] = processedContent — OK (processMessageContent returns a string).
- transformVendor (line 361): attributes[SpanAttributes.LLM_SYSTEM] = mappedVendor || vendor — ensure vendor is string: mappedVendor || (typeof vendor === 'string' ? vendor : JSON.stringify(vendor)).
- transformToolCalls (lines 383, 386): span.attributes[SpanAttributes.TRACELOOP_ENTITY_INPUT/OUTPUT] = span.attributes["ai.toolCall.args"/"ai.toolCall.result"] — stringify these values (or conditionally stringify) before assigning.
🧹 Nitpick comments (2)
packages/traceloop-sdk/src/lib/tracing/span-processor.ts (1)
162-164: Gate name transforms to AI spans to avoid accidental renames.Only rename when the instrumentation is "ai".
Apply this diff:
- transformAiSdkSpanNames(span); + const scopeName = + (span as any).instrumentationScope?.name || + (span as any).instrumentationLibrary?.name; + if (scopeName === "ai") { + transformAiSdkSpanNames(span); + }packages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts (1)
378-390: Map tool-call args/result independently (don’t require both).Today nothing is mapped if only args or only result exist.
Apply this diff:
-const transformToolCalls = (span: ReadableSpan): void => { - if ( - span.attributes["ai.toolCall.args"] && - span.attributes["ai.toolCall.result"] - ) { - span.attributes[SpanAttributes.TRACELOOP_ENTITY_INPUT] = - span.attributes["ai.toolCall.args"]; - delete span.attributes["ai.toolCall.args"]; - span.attributes[SpanAttributes.TRACELOOP_ENTITY_OUTPUT] = - span.attributes["ai.toolCall.result"]; - delete span.attributes["ai.toolCall.result"]; - } -}; +const transformToolCalls = (span: ReadableSpan): void => { + if (span.attributes["ai.toolCall.args"]) { + span.attributes[SpanAttributes.TRACELOOP_ENTITY_INPUT] = + span.attributes["ai.toolCall.args"]; + delete span.attributes["ai.toolCall.args"]; + } + if (span.attributes["ai.toolCall.result"]) { + span.attributes[SpanAttributes.TRACELOOP_ENTITY_OUTPUT] = + span.attributes["ai.toolCall.result"]; + delete span.attributes["ai.toolCall.result"]; + } +};
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
Disabled knowledge base sources:
- Linear integration is disabled by default for public repositories
You can enable these sources in your CodeRabbit configuration.
📒 Files selected for processing (2)
packages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts(3 hunks)packages/traceloop-sdk/src/lib/tracing/span-processor.ts(4 hunks)
🧰 Additional context used
📓 Path-based instructions (2)
packages/{instrumentation-*,traceloop-sdk}/**/*.{ts,tsx}
📄 CodeRabbit inference engine (CLAUDE.md)
Import AI/LLM semantic attribute constants from @traceloop/ai-semantic-conventions rather than hardcoding strings
Files:
packages/traceloop-sdk/src/lib/tracing/span-processor.tspackages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts
packages/traceloop-sdk/**/*.{ts,tsx}
📄 CodeRabbit inference engine (CLAUDE.md)
packages/traceloop-sdk/**/*.{ts,tsx}: Use the provided decorators (@workflow, @task, @agent) for workflow/task/agent spans instead of re-implementing them
For manual LLM operations, use trace.withLLMSpan from @traceloop/node-server-sdk
Files:
packages/traceloop-sdk/src/lib/tracing/span-processor.tspackages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts
🧠 Learnings (5)
📚 Learning: 2025-08-24T22:08:07.023Z
Learnt from: CR
PR: traceloop/openllmetry-js#0
File: CLAUDE.md:0-0
Timestamp: 2025-08-24T22:08:07.023Z
Learning: Applies to packages/traceloop-sdk/**/*.{ts,tsx} : Use the provided decorators (workflow, task, agent) for workflow/task/agent spans instead of re-implementing them
Applied to files:
packages/traceloop-sdk/src/lib/tracing/span-processor.tspackages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts
📚 Learning: 2025-08-24T22:08:07.023Z
Learnt from: CR
PR: traceloop/openllmetry-js#0
File: CLAUDE.md:0-0
Timestamp: 2025-08-24T22:08:07.023Z
Learning: Applies to packages/{instrumentation-*,traceloop-sdk}/**/*.{ts,tsx} : Import AI/LLM semantic attribute constants from traceloop/ai-semantic-conventions rather than hardcoding strings
Applied to files:
packages/traceloop-sdk/src/lib/tracing/span-processor.tspackages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts
📚 Learning: 2025-08-24T22:08:07.023Z
Learnt from: CR
PR: traceloop/openllmetry-js#0
File: CLAUDE.md:0-0
Timestamp: 2025-08-24T22:08:07.023Z
Learning: Applies to packages/instrumentation-*/**/*.{ts,tsx} : Instrumentations must create spans with appropriate AI/LLM semantic attributes for calls they wrap
Applied to files:
packages/traceloop-sdk/src/lib/tracing/span-processor.tspackages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts
📚 Learning: 2025-08-24T22:08:07.023Z
Learnt from: CR
PR: traceloop/openllmetry-js#0
File: CLAUDE.md:0-0
Timestamp: 2025-08-24T22:08:07.023Z
Learning: Applies to packages/traceloop-sdk/**/*.{ts,tsx} : For manual LLM operations, use trace.withLLMSpan from traceloop/node-server-sdk
Applied to files:
packages/traceloop-sdk/src/lib/tracing/span-processor.tspackages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts
📚 Learning: 2025-08-24T22:08:07.023Z
Learnt from: CR
PR: traceloop/openllmetry-js#0
File: CLAUDE.md:0-0
Timestamp: 2025-08-24T22:08:07.023Z
Learning: Applies to packages/ai-semantic-conventions/src/SemanticAttributes.ts : Define all AI/LLM span attribute constants in packages/ai-semantic-conventions/src/SemanticAttributes.ts
Applied to files:
packages/traceloop-sdk/src/lib/tracing/span-processor.tspackages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts
🧬 Code graph analysis (2)
packages/traceloop-sdk/src/lib/tracing/span-processor.ts (1)
packages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts (2)
transformAiSdkSpanNames(396-403)transformAiSdkSpanAttributes(405-411)
packages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts (1)
packages/ai-semantic-conventions/src/SemanticAttributes.ts (1)
SpanAttributes(17-61)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: Build and test
🔇 Additional comments (8)
packages/traceloop-sdk/src/lib/tracing/span-processor.ts (3)
5-7: Importing SDK Span is appropriate for updateName usage.Correct to use Span from @opentelemetry/sdk-trace-node here.
228-233: Good placement: attribute transforms on end before export.Running transformAiSdkSpanAttributes(span) before ensureSpanCompatibility is the right order.
5-7: Verify OpenTelemetry package versions align with SpanProcessor imports.Automated scan found no @opentelemetry/* entries in package.json files; packages/traceloop-sdk/src/lib/tracing/span-processor.ts imports @opentelemetry/sdk-trace-node — confirm where OTel deps are declared (monorepo root, peerDependencies, or left to consumers) and ensure the installed major version (v1 vs v2) is compatible with the SpanProcessor typings.
packages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts (5)
1-1: Correct: use SDK Span for updateName.updateName exists on SDK Span; this import is right.
4-13: Span-name mapping looks good.New names match the intended actions (run.ai/text.generate/object.generate/text.stream).
15-16: Tool-call span constant introduced appropriately.Clear dedicated key for tool-call handling.
405-411: Good split: attributes handled on end; names on start.Separation keeps span names stable while allowing attribute transforms once all data is present.
346-364: Vendor mapping: use OpenTelemetry gen_ai.system canonical identifiers (lowercase/dotted).Map VENDOR_MAPPING prefixes to the OTel GenAI canonical tokens (examples: openai, anthropic, gcp.vertex_ai, gcp.gemini, azure.ai.openai, aws.bedrock, cohere). For unknown/custom models use a friendly name or set gen_ai.system to _OTHER. Ensure transformVendor writes the canonical token to SpanAttributes.LLM_SYSTEM (not display names like "OpenAI" or "Google"). File: packages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts (transformVendor).
| const shouldHandleSpan = (span: ReadableSpan): boolean => { | ||
| return span.name in HANDLED_SPAN_NAMES; | ||
| return span.instrumentationScope.name === "ai"; | ||
| }; | ||
|
|
There was a problem hiding this comment.
Handle both instrumentationScope and instrumentationLibrary (OTel v1/v2).
Current check misses v1 where only instrumentationLibrary exists, causing transforms to be skipped.
Apply this diff:
-const shouldHandleSpan = (span: ReadableSpan): boolean => {
- return span.instrumentationScope.name === "ai";
-};
+const shouldHandleSpan = (span: ReadableSpan): boolean => {
+ const scopeName =
+ (span as any).instrumentationScope?.name ||
+ (span as any).instrumentationLibrary?.name;
+ return scopeName === "ai";
+};📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| const shouldHandleSpan = (span: ReadableSpan): boolean => { | |
| return span.name in HANDLED_SPAN_NAMES; | |
| return span.instrumentationScope.name === "ai"; | |
| }; | |
| const shouldHandleSpan = (span: ReadableSpan): boolean => { | |
| const scopeName = | |
| (span as any).instrumentationScope?.name || | |
| (span as any).instrumentationLibrary?.name; | |
| return scopeName === "ai"; | |
| }; |
🤖 Prompt for AI Agents
In packages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts around lines
392 to 395, the shouldHandleSpan function only checks
span.instrumentationScope.name (OTel v2) and therefore misses v1 spans that use
instrumentationLibrary; update the predicate to read the name from either
instrumentationScope or instrumentationLibrary (e.g. use
instrumentationScope?.name ?? instrumentationLibrary?.name) and compare that
value to "ai" so spans from both OTel v1 and v2 are handled.
| export const transformAiSdkSpanNames = (span: Span): void => { | ||
| if (span.name === TOOL_SPAN_NAME) { | ||
| span.updateName(`${span.attributes["ai.toolCall.name"] as string}.tool`); | ||
| } | ||
| if (span.name in HANDLED_SPAN_NAMES) { | ||
| span.updateName(HANDLED_SPAN_NAMES[span.name]); | ||
| } | ||
| }; |
There was a problem hiding this comment.
Guard against missing ai.toolCall.name to prevent 'undefined.tool'.
Only rename tool-call spans when the name attribute exists and is non-empty.
Apply this diff:
export const transformAiSdkSpanNames = (span: Span): void => {
- if (span.name === TOOL_SPAN_NAME) {
- span.updateName(`${span.attributes["ai.toolCall.name"] as string}.tool`);
- }
+ if (span.name === TOOL_SPAN_NAME) {
+ const toolName = (span.attributes["ai.toolCall.name"] as unknown) as string | undefined;
+ if (typeof toolName === "string" && toolName.length > 0) {
+ span.updateName(`${toolName}.tool`);
+ }
+ }
if (span.name in HANDLED_SPAN_NAMES) {
span.updateName(HANDLED_SPAN_NAMES[span.name]);
}
};📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| export const transformAiSdkSpanNames = (span: Span): void => { | |
| if (span.name === TOOL_SPAN_NAME) { | |
| span.updateName(`${span.attributes["ai.toolCall.name"] as string}.tool`); | |
| } | |
| if (span.name in HANDLED_SPAN_NAMES) { | |
| span.updateName(HANDLED_SPAN_NAMES[span.name]); | |
| } | |
| }; | |
| export const transformAiSdkSpanNames = (span: Span): void => { | |
| if (span.name === TOOL_SPAN_NAME) { | |
| const toolName = (span.attributes["ai.toolCall.name"] as unknown) as string | undefined; | |
| if (typeof toolName === "string" && toolName.length > 0) { | |
| span.updateName(`${toolName}.tool`); | |
| } | |
| } | |
| if (span.name in HANDLED_SPAN_NAMES) { | |
| span.updateName(HANDLED_SPAN_NAMES[span.name]); | |
| } | |
| }; |
🤖 Prompt for AI Agents
In packages/traceloop-sdk/src/lib/tracing/ai-sdk-transformations.ts around lines
396 to 403, the tool-call renaming can produce "undefined.tool" when
ai.toolCall.name is missing; change the logic so you only call
span.updateName(`${...}.tool`) if span.attributes["ai.toolCall.name"] is a
non-empty string (e.g. typeof === "string" &&
span.attributes["ai.toolCall.name"].trim() !== ""), otherwise skip that rename;
keep the subsequent HANDLED_SPAN_NAMES check as-is so other mappings still
apply.
| export const transformAiSdkSpanAttributes = (span: ReadableSpan): void => { | ||
| if (!shouldHandleSpan(span)) { | ||
| return; | ||
| } | ||
| transformAiSdkSpanName(span); | ||
| transformAiSdkAttributes(span.attributes); | ||
| transformLLMSpans(span.attributes); | ||
| transformToolCalls(span); |
There was a problem hiding this comment.
[STYLE] naming & args aren't consistent, if you changed transformAiSdkSpan -> transformAiSdkSpanAttributes than the methods used inside should also be with attrs suffix imo, and have the same signature
| span.updateName(`${span.attributes["ai.toolCall.name"] as string}.tool`); | ||
| } | ||
| if (span.name in HANDLED_SPAN_NAMES) { | ||
| span.updateName(HANDLED_SPAN_NAMES[span.name]); |
There was a problem hiding this comment.
nice! this updateName is a new api i think so. if that's the case this change will break the otel-v1 tho if we merge it.
just read about it, it exists since v1, makes me wonder why didn't we use it.
| [AI_GENERATE_TEXT]: "run.ai", | ||
| [AI_GENERATE_TEXT_DO_GENERATE]: "text.generate", | ||
| [AI_GENERATE_OBJECT_DO_GENERATE]: "object.generate", | ||
| [AI_STREAM_TEXT_DO_STREAM]: "text.stream", |
There was a problem hiding this comment.
these naming changes may break mor's monitors if merged.
There was a problem hiding this comment.
@galkleinman the name changes didn't work at all apparently
There was a problem hiding this comment.
Important
Looks good to me! 👍
Reviewed 08f5484 in 1 minute and 29 seconds. Click for details.
- Reviewed
940lines of code in4files - Skipped
0files when reviewing. - Skipped posting
8draft comments. View those below. - Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.
1. packages/traceloop-sdk/test/ai-sdk-transformations.test.ts:395
- Draft comment:
Excellent test for handling invalid JSON in 'ai.prompt.messages'; the attributes remain unchanged on parse failure. - Reason this comment was not posted:
Comment was not on a location in the diff, so it can't be submitted as a review comment.
2. packages/traceloop-sdk/test/ai-sdk-transformations.test.ts:428
- Draft comment:
The unescaping of JSON escape sequences in prompt messages is correctly validated. - Reason this comment was not posted:
Comment was not on a location in the diff, so it can't be submitted as a review comment.
3. packages/traceloop-sdk/test/ai-sdk-transformations.test.ts:888
- Draft comment:
Token usage calculations correctly handle both numeric and string values for prompt and completion tokens. - Reason this comment was not posted:
Comment did not seem useful. Confidence is useful =0%<= threshold50%This comment is purely informative, stating that token usage calculations handle both numeric and string values. It does not provide a suggestion, ask for confirmation, or point out a potential issue. It violates the rule against making purely informative comments.
4. packages/traceloop-sdk/test/decorators.test.ts:189
- Draft comment:
Decorator tests effectively verify dynamic workflow and task naming along with association property propagation. - Reason this comment was not posted:
Comment was not on a location in the diff, so it can't be submitted as a review comment.
5. packages/traceloop-sdk/test/decorators.test.ts:525
- Draft comment:
The parallel workflow test confirms that association properties are isolated between concurrent traces. - Reason this comment was not posted:
Comment was not on a location in the diff, so it can't be submitted as a review comment.
6. packages/traceloop-sdk/test/decorators.test.ts:640
- Draft comment:
Vercel AI span fix test properly aligns the span format with OpenLLMetry expectations, including model and token usage attributes. - Reason this comment was not posted:
Comment was not on a location in the diff, so it can't be submitted as a review comment.
7. packages/traceloop-sdk/test/decorators.test.ts:161
- Draft comment:
The suppressed workflow test verifies that no spans are created when 'suppressTracing' is enabled. - Reason this comment was not posted:
Comment was not on a location in the diff, so it can't be submitted as a review comment.
8. packages/traceloop-sdk/test/ai-sdk-transformations.test.ts:5
- Draft comment:
Typographical Error: The constant nameASSOCATION_PROPERTIES_KEYseems misspelled. It likely should beASSOCIATION_PROPERTIES_KEY. - Reason this comment was not posted:
Decided after close inspection that this draft comment was likely wrong and/or not actionable: usefulness confidence = 0% vs. threshold = 50% The comment points out a legitimate spelling error in a constant name that was imported. However, this is an import statement - we can't see the actual constant definition in this diff. The typo exists in another file that we can't see. According to the rules, we should "ignore cross-file issues" and "only think about the file you are reviewing". The typo seems real and should be fixed, but fixing it would require changes in another file where the constant is defined. Am I being too strict with the cross-file rule? No, the rules are clear that we should ignore cross-file issues. Even though the typo is real, fixing it requires changes to another file where the constant is defined. We can't properly review this without seeing that file. Delete the comment since it requires changes in another file that we can't see in this diff. The cross-file rule takes precedence.
Workflow ID: wflow_vQl2ibH6AaMvPSkZ
You can customize
There was a problem hiding this comment.
Important
Looks good to me! 👍
Reviewed c6827a0 in 36 seconds. Click for details.
- Reviewed
169lines of code in2files - Skipped
0files when reviewing. - Skipped posting
2draft comments. View those below. - Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.
1. packages/traceloop-sdk/test/ai-sdk-integration.test.ts:233
- Draft comment:
Refactored the arrow function in the.findcall by removing the trailing comma. This concise inline format improves clarity without changing functionality. - Reason this comment was not posted:
Confidence changes required:0%<= threshold50%None
2. packages/traceloop-sdk/test/ai-sdk-transformations.test.ts:1555
- Draft comment:
The assert statements were reformatted (e.g. multi-line assertions in telemetry metadata and other test blocks) for greater consistency and readability. No functional changes were introduced. - Reason this comment was not posted:
Confidence changes required:0%<= threshold50%None
Workflow ID: wflow_dg3vMM9GFxMlH9cN
You can customize
2 participants
Important
Enhances Vercel AI SDK tool call handling with standardized span names, improved attribute mapping, and refactored span processing logic.
ai-sdk-transformations.ts.span-processor.ts.ai-sdk-integration.test.ts,ai-sdk-transformations.test.ts, anddecorators.test.tsto verify new span names and attribute transformations.This description was created by
for c6827a0. You can customize this summary. It will automatically update as commits are pushed.
Summary by CodeRabbit
New Features
Refactor
Tests