Skip to content

feat(gen_ai): add new Gen AI attributes #221

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
obostjancic merged 7 commits into from
Jan 20, 2026
Merged

feat(gen_ai): add new Gen AI attributes #221

Show file tree
Hide file tree
Changes from 6 commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
3b20cf3
feat(gen_ai): add new Gen AI attributes
obostjancic Jan 19, 2026
45d6085
test fix
obostjancic Jan 19, 2026
57c30a8
reformat
obostjancic Jan 19, 2026
c178fd2
fix is_in_otel
obostjancic Jan 19, 2026
c48647e
changelog
obostjancic Jan 19, 2026
c0a9adf
changelog fix
obostjancic Jan 19, 2026
da28d7e
change input messages schema
obostjancic Jan 20, 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
4 changes: 2 additions & 2 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,7 @@
- feat(ai): Add gen_ai.usage.input_tokens.cache_write ([#217](https://github.com/getsentry/sentry-conventions/pull/217))
- feat(attributes): Add sentry.normalized_db_query.hash ([#200](https://github.com/getsentry/sentry-conventions/pull/200))
- feat(attributes): Add sentry.category attribute ([#218](https://github.com/getsentry/sentry-conventions/pull/218))
- Add new Gen AI attributes ([#221](https://github.com/getsentry/sentry-conventions/pull/221))

## 0.3.1

Expand Down Expand Up @@ -55,12 +56,11 @@
- feat(sentry): Add sentry.observed_timestamp_nanos ([#137](https://github.com/getsentry/sentry-conventions/pull/137))
- dynamic-sampling: add field conventions for dynamic sampling context ([#128](https://github.com/getsentry/sentry-conventions/pull/128))
- chore(ai): Clean up of `sentry._internal.segment.contains_gen_ai_spans` ([#155](https://github.com/getsentry/sentry-conventions/pull/155))
- feat(attributes): Add sentry._internal.replay_is_buffering ([#159](https://github.com/getsentry/sentry-conventions/pull/159))
- feat(attributes): Add sentry.\_internal.replay_is_buffering ([#159](https://github.com/getsentry/sentry-conventions/pull/159))
- feat: Add vercel log drain attributes ([#163](https://github.com/getsentry/sentry-conventions/pull/163))
- feat(attributes) add MCP related attributes ([#164](https://github.com/getsentry/sentry-conventions/pull/164))
- feat(attributes): Add MDC log attributes ([#167](https://github.com/getsentry/sentry-conventions/pull/167))


### Fixes

- fix(name): Remove duplicate GraphQL op ([#152](https://github.com/getsentry/sentry-conventions/pull/152))
Expand Down
18 changes: 12 additions & 6 deletions generated/attributes/all.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@

This page lists all available attributes across all categories.

Total attributes: 415
Total attributes: 421

## Stable Attributes

Expand Down Expand Up @@ -81,13 +81,13 @@ Total attributes: 415
| [`gen_ai.cost.output_tokens`](./gen_ai.md#gen_aicostoutput_tokens) | The cost of tokens used for creating the AI output in USD (without reasoning tokens). |
| [`gen_ai.cost.total_tokens`](./gen_ai.md#gen_aicosttotal_tokens) | The total cost for the tokens used. |
| [`gen_ai.embeddings.input`](./gen_ai.md#gen_aiembeddingsinput) | The input to the embeddings model. |
| [`gen_ai.input.messages`](./gen_ai.md#gen_aiinputmessages) | The messages passed to the model. It has to be a stringified version of an array of objects. The `role` attribute of each object must be `"user"`, `"assistant"`, `"tool"`, or `"system"`. For messages of the role `"tool"`, the `content` can be a string or an arbitrary object with information about the tool call. For other messages the `content` can be either a string or a list of objects in the format `{type: "text", text:"..."}`. |
| [`gen_ai.operation.name`](./gen_ai.md#gen_aioperationname) | The name of the operation being performed. |
| [`gen_ai.operation.type`](./gen_ai.md#gen_aioperationtype) | The type of AI operation. Must be one of 'agent', 'ai_client', 'tool', 'handoff', 'guardrail'. Makes querying for spans in the UI easier. |
| [`gen_ai.output.messages`](./gen_ai.md#gen_aioutputmessages) | The model's response messages. It has to be a stringified version of an array of message objects, which can include text responses and tool calls. |
| [`gen_ai.pipeline.name`](./gen_ai.md#gen_aipipelinename) | Name of the AI pipeline or chain being executed. |
| [`gen_ai.request.available_tools`](./gen_ai.md#gen_airequestavailable_tools) | The available tools for the model. It has to be a stringified version of an array of objects. |
| [`gen_ai.request.frequency_penalty`](./gen_ai.md#gen_airequestfrequency_penalty) | Used to reduce repetitiveness of generated tokens. The higher the value, the stronger a penalty is applied to previously present tokens, proportional to how many times they have already appeared in the prompt or prior generation. |
| [`gen_ai.request.max_tokens`](./gen_ai.md#gen_airequestmax_tokens) | The maximum number of tokens to generate in the response. |
| [`gen_ai.request.messages`](./gen_ai.md#gen_airequestmessages) | The messages passed to the model. It has to be a stringified version of an array of objects. The `role` attribute of each object must be `"user"`, `"assistant"`, `"tool"`, or `"system"`. For messages of the role `"tool"`, the `content` can be a string or an arbitrary object with information about the tool call. For other messages the `content` can be either a string or a list of objects in the format `{type: "text", text:"..."}`. |
| [`gen_ai.request.model`](./gen_ai.md#gen_airequestmodel) | The model identifier being used for the request. |
| [`gen_ai.request.presence_penalty`](./gen_ai.md#gen_airequestpresence_penalty) | Used to reduce repetitiveness of generated tokens. Similar to frequency_penalty, except that this penalty is applied equally to all tokens that have already appeared, regardless of their exact frequencies. |
| [`gen_ai.request.seed`](./gen_ai.md#gen_airequestseed) | The seed, ideally models given the same seed and same other parameters will produce the exact same output. |
Expand All @@ -98,11 +98,12 @@ Total attributes: 415
| [`gen_ai.response.id`](./gen_ai.md#gen_airesponseid) | Unique identifier for the completion. |
| [`gen_ai.response.model`](./gen_ai.md#gen_airesponsemodel) | The vendor-specific ID of the model used. |
| [`gen_ai.response.streaming`](./gen_ai.md#gen_airesponsestreaming) | Whether or not the AI model call's response was streamed back asynchronously |
| [`gen_ai.response.text`](./gen_ai.md#gen_airesponsetext) | The model's response text messages. It has to be a stringified version of an array of response text messages. |
| [`gen_ai.response.tokens_per_second`](./gen_ai.md#gen_airesponsetokens_per_second) | The total output tokens per seconds throughput |
| [`gen_ai.response.tool_calls`](./gen_ai.md#gen_airesponsetool_calls) | The tool calls in the model's response. It has to be a stringified version of an array of objects. |
| [`gen_ai.system`](./gen_ai.md#gen_aisystem) | The provider of the model. |
| [`gen_ai.system.message`](./gen_ai.md#gen_aisystemmessage) | The system instructions passed to the model. |
| [`gen_ai.system_instructions`](./gen_ai.md#gen_aisystem_instructions) | The system instructions passed to the model. |
| [`gen_ai.tool.call.arguments`](./gen_ai.md#gen_aitoolcallarguments) | The arguments of the tool call. It has to be a stringified version of the arguments to the tool. |
| [`gen_ai.tool.call.result`](./gen_ai.md#gen_aitoolcallresult) | The result of the tool call. It has to be a stringified version of the result of the tool. |
| [`gen_ai.tool.definitions`](./gen_ai.md#gen_aitooldefinitions) | The list of source system tool definitions available to the GenAI agent or model. |
| [`gen_ai.tool.description`](./gen_ai.md#gen_aitooldescription) | The description of the tool being used. |
| [`gen_ai.tool.input`](./gen_ai.md#gen_aitoolinput) | The input of the tool being used. It has to be a stringified version of the input to the tool. |
| [`gen_ai.tool.message`](./gen_ai.md#gen_aitoolmessage) | The response from a tool or function call passed to the model. |
Expand Down Expand Up @@ -389,6 +390,11 @@ Total attributes: 415
| [`environment`](./general.md#environment) | [`sentry.environment`](./sentry.md#sentryenvironment) |
| [`fs_error`](./general.md#fs_error) | [`error.type`](./error.md#errortype) |
| [`gen_ai.prompt`](./gen_ai.md#gen_aiprompt) | No replacement |
| [`gen_ai.request.available_tools`](./gen_ai.md#gen_airequestavailable_tools) | [`gen_ai.tool.definitions`](./gen_ai.md#gen_aitooldefinitions) |
| [`gen_ai.request.messages`](./gen_ai.md#gen_airequestmessages) | [`gen_ai.input.messages`](./gen_ai.md#gen_aiinputmessages) |
| [`gen_ai.response.text`](./gen_ai.md#gen_airesponsetext) | [`gen_ai.output.messages`](./gen_ai.md#gen_aioutputmessages) |
| [`gen_ai.response.tool_calls`](./gen_ai.md#gen_airesponsetool_calls) | [`gen_ai.output.messages`](./gen_ai.md#gen_aioutputmessages) |
| [`gen_ai.system.message`](./gen_ai.md#gen_aisystemmessage) | [`gen_ai.system_instructions`](./gen_ai.md#gen_aisystem_instructions) |
| [`gen_ai.usage.completion_tokens`](./gen_ai.md#gen_aiusagecompletion_tokens) | [`gen_ai.usage.output_tokens`](./gen_ai.md#gen_aiusageoutput_tokens) |
| [`gen_ai.usage.prompt_tokens`](./gen_ai.md#gen_aiusageprompt_tokens) | [`gen_ai.usage.input_tokens`](./gen_ai.md#gen_aiusageinput_tokens) |
| [`http.client_ip`](./http.md#httpclient_ip) | [`client.address`](./client.md#clientaddress) |
Expand Down
173 changes: 125 additions & 48 deletions generated/attributes/gen_ai.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,13 +10,13 @@
- [gen_ai.cost.output_tokens](#gen_aicostoutput_tokens)
- [gen_ai.cost.total_tokens](#gen_aicosttotal_tokens)
- [gen_ai.embeddings.input](#gen_aiembeddingsinput)
- [gen_ai.input.messages](#gen_aiinputmessages)
- [gen_ai.operation.name](#gen_aioperationname)
- [gen_ai.operation.type](#gen_aioperationtype)
- [gen_ai.output.messages](#gen_aioutputmessages)
- [gen_ai.pipeline.name](#gen_aipipelinename)
- [gen_ai.request.available_tools](#gen_airequestavailable_tools)
- [gen_ai.request.frequency_penalty](#gen_airequestfrequency_penalty)
- [gen_ai.request.max_tokens](#gen_airequestmax_tokens)
- [gen_ai.request.messages](#gen_airequestmessages)
- [gen_ai.request.model](#gen_airequestmodel)
- [gen_ai.request.presence_penalty](#gen_airequestpresence_penalty)
- [gen_ai.request.seed](#gen_airequestseed)
Expand All @@ -27,11 +27,12 @@
- [gen_ai.response.id](#gen_airesponseid)
- [gen_ai.response.model](#gen_airesponsemodel)
- [gen_ai.response.streaming](#gen_airesponsestreaming)
- [gen_ai.response.text](#gen_airesponsetext)
- [gen_ai.response.tokens_per_second](#gen_airesponsetokens_per_second)
- [gen_ai.response.tool_calls](#gen_airesponsetool_calls)
- [gen_ai.system](#gen_aisystem)
- [gen_ai.system.message](#gen_aisystemmessage)
- [gen_ai.system_instructions](#gen_aisystem_instructions)
- [gen_ai.tool.call.arguments](#gen_aitoolcallarguments)
- [gen_ai.tool.call.result](#gen_aitoolcallresult)
- [gen_ai.tool.definitions](#gen_aitooldefinitions)
- [gen_ai.tool.description](#gen_aitooldescription)
- [gen_ai.tool.input](#gen_aitoolinput)
- [gen_ai.tool.message](#gen_aitoolmessage)
Expand All @@ -47,6 +48,11 @@
- [gen_ai.user.message](#gen_aiusermessage)
- [Deprecated Attributes](#deprecated-attributes)
- [gen_ai.prompt](#gen_aiprompt)
- [gen_ai.request.available_tools](#gen_airequestavailable_tools)
- [gen_ai.request.messages](#gen_airequestmessages)
- [gen_ai.response.text](#gen_airesponsetext)
- [gen_ai.response.tool_calls](#gen_airesponsetool_calls)
- [gen_ai.system.message](#gen_aisystemmessage)
- [gen_ai.usage.completion_tokens](#gen_aiusagecompletion_tokens)
- [gen_ai.usage.prompt_tokens](#gen_aiusageprompt_tokens)

Expand Down Expand Up @@ -129,6 +135,17 @@ The input to the embeddings model.
| Exists in OpenTelemetry | No |
| Example | `What's the weather in Paris?` |

### gen_ai.input.messages

The messages passed to the model. It has to be a stringified version of an array of objects. The `role` attribute of each object must be `"user"`, `"assistant"`, `"tool"`, or `"system"`. For messages of the role `"tool"`, the `content` can be a string or an arbitrary object with information about the tool call. For other messages the `content` can be either a string or a list of objects in the format `{type: "text", text:"..."}`.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | Yes |
| Example | `[{"role": "user", "parts": [{"type": "text", "content": "What is the weather in Paris?"}]}]` |

### gen_ai.operation.name

The name of the operation being performed.
Expand All @@ -151,28 +168,28 @@ The type of AI operation. Must be one of 'agent', 'ai_client', 'tool', 'handoff'
| Exists in OpenTelemetry | No |
| Example | `tool` |

### gen_ai.pipeline.name
### gen_ai.output.messages

Name of the AI pipeline or chain being executed.
The model's response messages. It has to be a stringified version of an array of message objects, which can include text responses and tool calls.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | No |
| Example | `Autofix Pipeline` |
| Aliases | `ai.pipeline.name` |
| Exists in OpenTelemetry | Yes |
| Example | `[{"role": "assistant", "parts": [{"type": "text", "content": "The weather in Paris is currently rainy with a temperature of 57°F."}], "finish_reason": "stop"}]` |

### gen_ai.request.available_tools
### gen_ai.pipeline.name

The available tools for the model. It has to be a stringified version of an array of objects.
Name of the AI pipeline or chain being executed.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | No |
| Example | `[{"name": "get_weather", "description": "Get the weather for a given location"}, {"name": "get_news", "description": "Get the news for a given topic"}]` |
| Example | `Autofix Pipeline` |
| Aliases | `ai.pipeline.name` |

### gen_ai.request.frequency_penalty

Expand All @@ -197,18 +214,6 @@ The maximum number of tokens to generate in the response.
| Exists in OpenTelemetry | Yes |
| Example | `2048` |

### gen_ai.request.messages

The messages passed to the model. It has to be a stringified version of an array of objects. The `role` attribute of each object must be `"user"`, `"assistant"`, `"tool"`, or `"system"`. For messages of the role `"tool"`, the `content` can be a string or an arbitrary object with information about the tool call. For other messages the `content` can be either a string or a list of objects in the format `{type: "text", text:"..."}`.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | No |
| Example | `[{"role": "system", "content": "Generate a random number."}, {"role": "user", "content": [{"text": "Generate a random number between 0 and 10.", "type": "text"}]}, {"role": "tool", "content": {"toolCallId": "1", "toolName": "Weather", "output": "rainy"}}]` |
| Aliases | `ai.input_messages` |

### gen_ai.request.model

The model identifier being used for the request.
Expand Down Expand Up @@ -328,61 +333,72 @@ Whether or not the AI model call's response was streamed back asynchronously
| Example | `true` |
| Aliases | `ai.streaming` |

### gen_ai.response.text
### gen_ai.response.tokens_per_second

The model's response text messages. It has to be a stringified version of an array of response text messages.
The total output tokens per seconds throughput

| Property | Value |
| --- | --- |
| Type | `double` |
| Has PII | false |
| Exists in OpenTelemetry | No |
| Example | `12345.67` |

### gen_ai.system

The provider of the model.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | No |
| Example | `["The weather in Paris is rainy and overcast, with temperatures around 57°F", "The weather in London is sunny and warm, with temperatures around 65°F"]` |
| Exists in OpenTelemetry | Yes |
| Example | `openai` |
| Aliases | `ai.model.provider` |

### gen_ai.response.tokens_per_second
### gen_ai.system_instructions

The total output tokens per seconds throughput
The system instructions passed to the model.

| Property | Value |
| --- | --- |
| Type | `double` |
| Has PII | false |
| Exists in OpenTelemetry | No |
| Example | `12345.67` |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | Yes |
| Example | `You are a helpful assistant` |

### gen_ai.response.tool_calls
### gen_ai.tool.call.arguments

The tool calls in the model's response. It has to be a stringified version of an array of objects.
The arguments of the tool call. It has to be a stringified version of the arguments to the tool.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | No |
| Example | `[{"name": "get_weather", "arguments": {"location": "Paris"}}]` |
| Exists in OpenTelemetry | Yes |
| Example | `{"location": "Paris"}` |

### gen_ai.system
### gen_ai.tool.call.result

The provider of the model.
The result of the tool call. It has to be a stringified version of the result of the tool.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | Yes |
| Example | `openai` |
| Aliases | `ai.model.provider` |
| Example | `rainy, 57°F` |

### gen_ai.system.message
### gen_ai.tool.definitions

The system instructions passed to the model.
The list of source system tool definitions available to the GenAI agent or model.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | true |
| Exists in OpenTelemetry | No |
| Example | `You are a helpful assistant` |
| Has PII | maybe |
| Exists in OpenTelemetry | Yes |
| Example | `[{"type": "function", "name": "get_current_weather", "description": "Get the current weather in a given location", "parameters": {"type": "object", "properties": {"location": {"type": "string", "description": "The city and state, e.g. San Francisco, CA"}, "unit": {"type": "string", "enum": ["celsius", "fahrenheit"]}}, "required": ["location", "unit"]}}]` |

### gen_ai.tool.description

Expand Down Expand Up @@ -548,6 +564,67 @@ The input messages sent to the model
| Deprecated | Yes, no replacement at this time |
| Deprecation Reason | Deprecated from OTEL, use gen_ai.input.messages with the new format instead. |

### gen_ai.request.available_tools

The available tools for the model. It has to be a stringified version of an array of objects.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | No |
| Example | `[{"name": "get_weather", "description": "Get the weather for a given location"}, {"name": "get_news", "description": "Get the news for a given topic"}]` |
| Deprecated | Yes, use `gen_ai.tool.definitions` instead |

### gen_ai.request.messages

The messages passed to the model. It has to be a stringified version of an array of objects. The `role` attribute of each object must be `"user"`, `"assistant"`, `"tool"`, or `"system"`. For messages of the role `"tool"`, the `content` can be a string or an arbitrary object with information about the tool call. For other messages the `content` can be either a string or a list of objects in the format `{type: "text", text:"..."}`.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | No |
| Example | `[{"role": "system", "content": "Generate a random number."}, {"role": "user", "content": [{"text": "Generate a random number between 0 and 10.", "type": "text"}]}, {"role": "tool", "content": {"toolCallId": "1", "toolName": "Weather", "output": "rainy"}}]` |
| Deprecated | Yes, use `gen_ai.input.messages` instead |
| Aliases | `ai.input_messages` |

### gen_ai.response.text

The model's response text messages. It has to be a stringified version of an array of response text messages.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | No |
| Example | `["The weather in Paris is rainy and overcast, with temperatures around 57°F", "The weather in London is sunny and warm, with temperatures around 65°F"]` |
| Deprecated | Yes, use `gen_ai.output.messages` instead |

### gen_ai.response.tool_calls

The tool calls in the model's response. It has to be a stringified version of an array of objects.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | maybe |
| Exists in OpenTelemetry | No |
| Example | `[{"name": "get_weather", "arguments": {"location": "Paris"}}]` |
| Deprecated | Yes, use `gen_ai.output.messages` instead |

### gen_ai.system.message

The system instructions passed to the model.

| Property | Value |
| --- | --- |
| Type | `string` |
| Has PII | true |
| Exists in OpenTelemetry | No |
| Example | `You are a helpful assistant` |
| Deprecated | Yes, use `gen_ai.system_instructions` instead |

### gen_ai.usage.completion_tokens

The number of tokens used in the GenAI response (completion).
Expand Down
Loading
Loading