Attribute Keys
Reference for the judgment.* span attribute keys Judgment reads when ingesting traces.
Judgment reads a fixed set of judgment.* OpenTelemetry span attributes when it
ingests a trace. The Judgeval SDKs set these for you. You only need this
reference when you write attributes yourself — with
set_attribute,
from a raw OpenTelemetry SDK, or as the target key of an
attribute mapper.
Anything outside this list is still stored and searchable, but Judgment treats it as an arbitrary custom attribute.
Core
Control how a span is displayed in the trace view.
| Key | Description |
|---|---|
judgment.span_name | The name shown for this step in the trace. |
judgment.span_kind | The type of step (llm, tool, function, agent, ...). |
judgment.input | The input a step receives. |
judgment.output | The output a step produces. |
Identifiers
Group traces so they can be filtered and broken down in monitoring.
| Key | Description |
|---|---|
judgment.session_id | Ties multiple traces to one conversation or session. |
judgment.customer_id | Which of your customers (tenant or account) a trace belongs to. |
judgment.customer_user_id | Which individual end user a trace belongs to. |
Usage and cost
Judgment prices a span from its provider, model, and token counts. If you
already have a cost figure, set judgment.usage.total_cost_usd instead.
| Key | Description |
|---|---|
judgment.llm.provider | The model provider, e.g. openai or anthropic. |
judgment.llm.model | The exact model used, e.g. gpt-4o. |
judgment.usage.non_cached_input_tokens | Input tokens charged at the normal (uncached) rate. |
judgment.usage.cache_creation_input_tokens | Input tokens written into the prompt cache. |
judgment.usage.cache_read_input_tokens | Input tokens read from the prompt cache at the cheaper cached rate. |
judgment.usage.reasoning_tokens | Tokens a reasoning model spends thinking. |
judgment.usage.output_tokens | Tokens the model generated in its response. |
judgment.usage.non_cached_input_image_tokens | Image input tokens, for models that bill them separately. |
judgment.usage.output_image_tokens | Generated image tokens, for models that bill them separately. |
judgment.usage.total_cost_usd | A cost you computed yourself for this step, in USD. |