fix(openai): use SpanAttributes instead of GenAIAttributes for cache token attributes #3442

hayke102 · 2025-11-13T14:31:50Z

Summary

Fixed AttributeError by replacing GenAIAttributes with SpanAttributes for cache-related token attributes in Langchain and Anthropic instrumentation packages.

Problem

The code was incorrectly attempting to use GenAIAttributes.GEN_AI_USAGE_CACHE_READ_INPUT_TOKENS and GenAIAttributes.GEN_AI_USAGE_CACHE_CREATION_INPUT_TOKENS, but these attributes don't exist in the upstream OpenTelemetry incubating semantic conventions.

These cache token attributes are custom extensions added by OpenLLMetry to support prompt caching features (used by Anthropic, OpenAI, etc.) and are defined in the local SpanAttributes class in opentelemetry-semantic-conventions-ai package.

Solution

Replaced GenAIAttributes with SpanAttributes for both cache token attributes:

GEN_AI_USAGE_CACHE_READ_INPUT_TOKENS
GEN_AI_USAGE_CACHE_CREATION_INPUT_TOKENS

Files Changed

packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/span_utils.py
- Line 349: Fixed cache read tokens attribute
packages/opentelemetry-instrumentation-anthropic/opentelemetry/instrumentation/anthropic/__init__.py
- Lines 286, 290, 400, 404: Fixed both cache read and cache creation tokens attributes in sync and async functions

Impact

This bug would have caused AttributeError at runtime when processing responses that include cache token information (e.g., Anthropic's prompt caching, OpenAI's cached tokens).

Testing

The fix aligns with the pattern already used correctly in other instrumentation packages:

opentelemetry-instrumentation-openai (correctly uses SpanAttributes.LLM_USAGE_CACHE_READ_INPUT_TOKENS)
traceloop-sdk (correctly uses SpanAttributes for cache attributes)

🤖 Generated with Claude Code

Important

Fixes AttributeError by replacing GenAIAttributes with SpanAttributes for cache token attributes in Langchain and Anthropic packages.

Behavior:
- Replaced GenAIAttributes with SpanAttributes for cache token attributes GEN_AI_USAGE_CACHE_READ_INPUT_TOKENS and GEN_AI_USAGE_CACHE_CREATION_INPUT_TOKENS.
- Fixes AttributeError when processing cache token information in Langchain and Anthropic packages.
Files Changed:
- span_utils.py in Langchain: Line 349.
- __init__.py in Anthropic: Lines 286, 290, 400, 404.
Impact:
- Prevents runtime AttributeError in cache token processing.
Testing:
- Aligns with correct usage in opentelemetry-instrumentation-openai and traceloop-sdk.

^{This description was created by}^{for 1dc9943. You can customize this summary. It will automatically update as commits are pushed.}

Summary by CodeRabbit

Refactor
- Standardized cache-related token-usage telemetry across OpenTelemetry instrumentation to use the unified SpanAttributes namespace, ensuring consistent observability and attribute naming for cache read/creation token metrics.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

…tributes Fixed AttributeError by replacing GenAIAttributes with SpanAttributes for cache-related token attributes (GEN_AI_USAGE_CACHE_READ_INPUT_TOKENS and GEN_AI_USAGE_CACHE_CREATION_INPUT_TOKENS). These cache token attributes are custom extensions added by OpenLLMetry to support prompt caching features (Anthropic, OpenAI) and are not part of the upstream OpenTelemetry incubating semantic conventions. They are defined in the local SpanAttributes class, not in GenAIAttributes. Changed files: - opentelemetry-instrumentation-langchain/span_utils.py (line 349) - opentelemetry-instrumentation-anthropic/__init__.py (lines 286, 290, 400, 404) 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>

CLAassistant · 2025-11-13T14:32:04Z

All committers have signed the CLA.

coderabbitai · 2025-11-13T14:32:04Z

Walkthrough

This pull request replaces cache-related token attribute constants from GenAIAttributes to SpanAttributes in two instrumentation packages (Anthropic and LangChain). No control-flow or error-handling changes were made.

Changes

Cohort / File(s)	Summary
Anthropic instrumentation `packages/opentelemetry-instrumentation-anthropic/opentelemetry/instrumentation/anthropic/__init__.py`	Replaced `GenAIAttributes.GEN_AI_USAGE_CACHE_READ_INPUT_TOKENS` with `SpanAttributes.GEN_AI_USAGE_CACHE_READ_INPUT_TOKENS` and `GenAIAttributes.GEN_AI_USAGE_CACHE_CREATION_INPUT_TOKENS` with `SpanAttributes.GEN_AI_USAGE_CACHE_CREATION_INPUT_TOKENS` in `_aset_token_usage` and `_set_token_usage`.
LangChain instrumentation `packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/span_utils.py`	Replaced `GenAIAttributes.GEN_AI_USAGE_CACHE_READ_INPUT_TOKENS` with `SpanAttributes.GEN_AI_USAGE_CACHE_READ_INPUT_TOKENS` in `set_chat_response`.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Potential review focus:

Confirm correct import/source of SpanAttributes and that no other GenAIAttributes usages remain for these keys.
Verify unit tests or span-attribute consumers expect the new attribute names.

Possibly related PRs

fix(openai): add streaming support for responses.create() api #3437: Performs the same replacement of GenAIAttributes cache-related constants with SpanAttributes equivalents across instrumentation modules.

Poem

🐰 I hopped through constants, small and bright,
Swapping GenAI for Span in the soft moonlight.
Tokens now rest where names align,
Cache metrics tidy, neat, and fine.
🥕✨

Pre-merge checks and finishing touches

❌ Failed checks (2 warnings)

Check name	Status	Explanation	Resolution
Title check	⚠️ Warning	The title mentions 'openai' but the PR primarily fixes cache token attribute issues in Langchain and Anthropic instrumentation, not OpenAI.	Revise title to accurately reflect the packages being fixed, such as: 'fix: use SpanAttributes instead of GenAIAttributes for cache token attributes in anthropic and langchain'
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	You can run `@coderabbitai generate docstrings` to improve docstring coverage.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

✨ Finishing touches

📝 Generate docstrings

🧪 Generate unit tests (beta)

Create PR with unit tests
Post copyable unit tests in a comment

📜 Recent 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.

📥 Commits

Reviewing files that changed from the base of the PR and between 1dc9943 and 9f48bcb.

📒 Files selected for processing (1)

packages/opentelemetry-instrumentation-anthropic/opentelemetry/instrumentation/anthropic/__init__.py (2 hunks)

🚧 Files skipped from review as they are similar to previous changes (1)

packages/opentelemetry-instrumentation-anthropic/opentelemetry/instrumentation/anthropic/init.py

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

ellipsis-dev

Important

Looks good to me! 👍

Reviewed everything up to 1dc9943 in 58 seconds. Click for details.

Reviewed 45 lines of code in 2 files
Skipped 0 files when reviewing.
Skipped posting 3 draft 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/opentelemetry-instrumentation-anthropic/opentelemetry/instrumentation/anthropic/__init__.py:286

Draft comment:
Replaced GenAIAttributes with SpanAttributes for cache read tokens in the async token usage block. This ensures the correct custom attribute is used.
Reason this comment was not posted:
Comment did not seem useful. Confidence is useful = 0% <= threshold 50% This comment is purely informative, explaining a change that was made without asking for confirmation or suggesting improvements. It doesn't align with the rules for good comments, as it doesn't provide a specific suggestion or ask for confirmation of intent.

2. packages/opentelemetry-instrumentation-anthropic/opentelemetry/instrumentation/anthropic/__init__.py:400

Draft comment:
Switched cache creation tokens attribute to use SpanAttributes in the sync token usage block to avoid AttributeError.
Reason this comment was not posted:
Comment did not seem useful. Confidence is useful = 0% <= threshold 50% This comment seems to be purely informative, explaining a change that was made. It doesn't ask for any specific action or provide a suggestion for improvement. According to the rules, purely informative comments should be removed.

3. packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/span_utils.py:346

Draft comment:
Updated cache read tokens attribute to use SpanAttributes instead of GenAIAttributes, aligning with the custom semantic conventions.
Reason this comment was not posted:
Comment did not seem useful. Confidence is useful = 0% <= threshold 50% This comment is purely informative, describing a change made in the code without providing any actionable feedback or suggestions. It doesn't ask for confirmation or suggest improvements, nor does it point out any potential issues. Therefore, it violates the rule against making purely informative comments.

Workflow ID: wflow_uNTK4jwkKuzbZIl9

^{You can customize}^{by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}

hayke102 · 2025-11-13T14:40:37Z

@coderabbitai generate docstrings

coderabbitai · 2025-11-13T14:40:43Z

✅ Actions performed

Initiated docstring generation; will generate only if new commits exist.

coderabbitai · 2025-11-13T14:40:46Z

Note

Docstrings generation - SUCCESS
Generated docstrings for this pull request at #3443

@hayke102

Docstrings generation was requested by @hayke102. * #3442 (comment) The following files were modified: * `packages/opentelemetry-instrumentation-anthropic/opentelemetry/instrumentation/anthropic/__init__.py` * `packages/opentelemetry-instrumentation-langchain/opentelemetry/instrumentation/langchain/span_utils.py`

nirga

Thanks @hayke102!

ellipsis-dev bot reviewed Nov 13, 2025

View reviewed changes

coderabbitai bot mentioned this pull request Nov 13, 2025

📝 Add docstrings to fix/cache-token-attributes #3443

Open

nirga changed the title ~~fix: use SpanAttributes instead of GenAIAttributes for cache token attributes~~ fix(openai): use SpanAttributes instead of GenAIAttributes for cache token attributes Nov 23, 2025

Merge branch 'main' into fix/cache-token-attributes

9f48bcb

nirga approved these changes Nov 23, 2025

View reviewed changes

nirga merged commit d9a0d7d into traceloop:main Nov 23, 2025
9 checks passed

coderabbitai bot mentioned this pull request Nov 23, 2025

fix(langchain): allow configuration of metadata key prefix #3367

Merged

4 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(openai): use SpanAttributes instead of GenAIAttributes for cache token attributes #3442

fix(openai): use SpanAttributes instead of GenAIAttributes for cache token attributes #3442

Uh oh!

hayke102 commented Nov 13, 2025 •

edited by coderabbitai bot

Loading

Uh oh!

CLAassistant commented Nov 13, 2025 •

edited

Loading

Uh oh!

coderabbitai bot commented Nov 13, 2025 •

edited

Loading

Uh oh!

ellipsis-dev bot left a comment

Uh oh!

hayke102 commented Nov 13, 2025

Uh oh!

coderabbitai bot commented Nov 13, 2025

Uh oh!

coderabbitai bot commented Nov 13, 2025 •

edited

Loading

Uh oh!

nirga left a comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

fix(openai): use SpanAttributes instead of GenAIAttributes for cache token attributes #3442

fix(openai): use SpanAttributes instead of GenAIAttributes for cache token attributes #3442

Uh oh!

Conversation

hayke102 commented Nov 13, 2025 • edited by coderabbitai bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Problem

Solution

Files Changed

Impact

Testing

Summary by CodeRabbit

Uh oh!

CLAassistant commented Nov 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

coderabbitai bot commented Nov 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Walkthrough

Changes

Estimated code review effort

Possibly related PRs

Poem

Pre-merge checks and finishing touches

Uh oh!

ellipsis-dev bot left a comment

Choose a reason for hiding this comment

Uh oh!

hayke102 commented Nov 13, 2025

Uh oh!

coderabbitai bot commented Nov 13, 2025

Uh oh!

coderabbitai bot commented Nov 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

nirga left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

hayke102 commented Nov 13, 2025 •

edited by coderabbitai bot

Loading

CLAassistant commented Nov 13, 2025 •

edited

Loading

coderabbitai bot commented Nov 13, 2025 •

edited

Loading

coderabbitai bot commented Nov 13, 2025 •

edited

Loading