feat: Add prompt and completion event logging (OpenTelemetry semantic conventions)

[MrKrishnabhagat]

/claim #2456

This PR implements support for emitting OpenTelemetry-compliant events for prompt and completion logging.

✅ Adds:

• use_legacy_attributes config (default: true)
• emit_prompt_events and emit_completion_events (default: true)
• record_prompt_and_completion() utility function
• Integrated into: completion_wrappers.py, chat_wrappers.py, embeddings_wrappers.py

Fully backward compatible with legacy attributes.

---

Important

Add OpenTelemetry-compliant prompt and completion event logging with configurable options and backward compatibility.

• Behavior:
  • Adds OpenTelemetry-compliant event logging for prompt and completion events via record_prompt_and_completion() in utils.py.
  • Configurable via use_legacy_attributes, emit_prompt_events, and emit_completion_events in config.py.
  • Integrated into completion_wrappers.py, chat_wrappers.py, embeddings_wrappers.py.
• Config:
  • Adds use_legacy_attributes, emit_prompt_events, emit_completion_events to Config class in config.py.
• Utility Function:
  • Adds record_prompt_and_completion() in utils.py to emit events or set attributes based on config.

^{This description was created by for 502f1ab. It will automatically update as commits are pushed.}

[CLAassistant]

All committers have signed the CLA.

[ellipsis-dev]

👍 Looks good to me! Reviewed everything up to 502f1ab in 1 minute and 42 seconds

More details

• Looked at 45 lines of code in 2 files
• Skipped 0 files when reviewing.
• Skipped posting 6 drafted comments based on config settings.

1. packages/opentelemetry-instrumentation-openai/opentelemetry/instrumentation/openai/shared/config.py:12

• Draft comment:
  The lambda for upload_base64_image only returns a type hint (str) instead of actual logic. Confirm if this is intended or if a proper implementation is needed.
• Reason this comment was not posted:
  Comment was on unchanged code.

2. packages/opentelemetry-instrumentation-openai/opentelemetry/instrumentation/openai/utils.py:117

• Draft comment:
  Consider adding more detailed documentation on when events vs. legacy attributes are used, as the flag toggles might be unclear to users.
• Reason this comment was not posted:
  Confidence changes required: 33% <= threshold 50%
  None

3. packages/opentelemetry-instrumentation-openai/opentelemetry/instrumentation/openai/shared/config.py:5

• Draft comment:
  New config flags (use_legacy_attributes, emit_prompt_events, emit_completion_events) have been added. Please ensure these are clearly documented so users know when and how each mode is used, and whether they might ever be used concurrently.
• Reason this comment was not posted:
  Confidence changes required: 33% <= threshold 50%
  None

4. packages/opentelemetry-instrumentation-openai/opentelemetry/instrumentation/openai/shared/config.py:12

• Draft comment:
  The default lambda for 'upload_base64_image' returns the built-in type 'str' rather than a string instance. If this no-op stub is intentional, consider adding a clarifying comment or returning an empty string instead.
• Reason this comment was not posted:
  Comment was on unchanged code.

5. packages/opentelemetry-instrumentation-openai/opentelemetry/instrumentation/openai/utils.py:117

• Draft comment:
  In 'record_prompt_and_completion', consider documenting that if 'use_legacy_attributes' is True, prompt/completion events are not emitted, and think about potential issues if very long strings are passed (which might affect span sizes).
• Reason this comment was not posted:
  Confidence changes required: 33% <= threshold 50%
  None

6. packages/opentelemetry-instrumentation-openai/opentelemetry/instrumentation/openai/utils.py:152

• Draft comment:
  Typo alert: The log message on line 152 says "OpenLLMetry failed to trace in %s, error: %s". If this is meant to refer to OpenTelemetry, please update the spelling accordingly.
• 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.

Workflow ID: wflow_Bioy5iGopIpRLzva

---

You can customize Ellipsis with 👍 / 👎 feedback, review rules, user-specific overrides, quiet mode, and more.