> ## Documentation Index
> Fetch the complete documentation index at: https://composio-preview.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Braintrust

> Braintrust is an AI engineering platform for managing prompts, datasets, experiments, evaluations, observability, and production AI applications.

![Braintrust logo](https://mintcdn.com/composio-preview/PudKL42NVsy5O-ND/images/toolkits/braintrust.svg?fit=max\&auto=format\&n=PudKL42NVsy5O-ND\&q=85\&s=c32bc729852c1e44801f2577509ebb7d)

* **Category:** artificial intelligence
* **Auth:** API\_KEY
* **Composio-managed OAuth available?** N/A
* **Tools:** 17
* **Triggers:** 0
* **Slug:** `BRAINTRUST`
* **Version:** 20261030\_00

## Tools

> [accordion] Create Dataset
>
> **Slug:** `BRAINTRUST_CREATE_DATASET`
>
> Create a named dataset under a Braintrust project, or return the existing dataset unchanged when its name already exists in that project.
>
> #### Input Parameters
>
> **name** `string` *required*
>
> Dataset name, which must be unique within the project. Braintrust returns the existing dataset unchanged when this name already exists.
>
> **tags** `array`
>
> Tags to attach to the dataset.
>
> **metadata** `object`
>
> User-controlled dataset metadata as an object.
>
> **project\_id** `string` *required*
>
> UUID of the Braintrust project that will contain the dataset.
>
> **description** `string`
>
> Optional human-readable dataset description.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Create Experiment
>
> **Slug:** `BRAINTRUST_CREATE_EXPERIMENT`
>
> Create an experiment under a Braintrust project. When its name already exists, Braintrust returns the existing experiment unchanged unless ensure\_new is true, which creates an experiment with a unique name.
>
> #### Input Parameters
>
> **name** `string`
>
> Experiment name. Names are unique within a project unless ensure\_new is true.
>
> **tags** `array`
>
> Tags to attach to the experiment.
>
> **public** `boolean`
>
> Whether anyone inside or outside the organization can view the experiment.
>
> **metadata** `object`
>
> User-controlled experiment metadata as an object.
>
> **dataset\_id** `string`
>
> UUID of the dataset linked to this experiment, when applicable.
>
> **ensure\_new** `boolean`
>
> Generate a unique experiment name on a name conflict. When false, Braintrust returns the existing same-name experiment unchanged.
>
> **project\_id** `string` *required*
>
> UUID of the Braintrust project that will contain the experiment.
>
> **description** `string`
>
> Optional human-readable experiment description.
>
> **dataset\_version** `string`
>
> Version of the linked dataset used for reproducibility.
>
> **base\_experiment\_id** `string`
>
> UUID of the default baseline experiment used for comparisons.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Create Project
>
> **Slug:** `BRAINTRUST_CREATE_PROJECT`
>
> Create a Braintrust project, or return the existing project unchanged when its name already exists.
>
> #### Input Parameters
>
> **name** `string` *required*
>
> Project name. Braintrust returns an existing project unchanged when this name already exists, so use a unique name when a new project is required.
>
> **org\_name** `string`
>
> Organization name for the rare case where the API key's active data plane supports organization selection. This does not reliably switch organizations.
>
> **description** `string`
>
> Optional human-readable project description.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Create Prompt
>
> **Slug:** `BRAINTRUST_CREATE_PROMPT`
>
> Upsert a versioned prompt configuration under a Braintrust project without invoking it. The same project and slug reuse the prompt ID but create a new version that replaces the stored prompt configuration with the submitted configuration. MCP attachments and tool-function bindings are unsupported.
>
> #### Input Parameters
>
> **name** `string` *required*
>
> Human-readable prompt name.
>
> **slug** `string` *required*
>
> Prompt identity slug within the project. Reusing the same project and slug keeps the prompt ID but creates a new version that replaces the stored prompt configuration with the submitted configuration.
>
> **tags** `array`
>
> Tags to attach to the prompt.
>
> **project\_id** `string` *required*
>
> UUID of the Braintrust project that will contain the prompt.
>
> **description** `string`
>
> Optional human-readable prompt description.
>
> **prompt\_data** `object`
>
> Structured prompt block, model options, and parser configuration. MCP attachments and tool\_functions are rejected by this schema.
>
> **function\_type** `string ("llm" | "scorer" | "task" | "tool" | "custom_view" | "preprocessor" | "facet" | "classifier" | "tag" | "parameters" | "sandbox")`
>
> Braintrust function classification assigned to the prompt.
>
> **template\_format** `string ("mustache" | "nunjucks" | "none")`
>
> Template syntax to store inside prompt\_data.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Delete Project
>
> **Slug:** `BRAINTRUST_DELETE_PROJECT`
>
> Soft-delete a Braintrust project and return its deletion state.
>
> #### Input Parameters
>
> **project\_id** `string` *required*
>
> UUID of the Braintrust project to soft-delete.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Delete Experiment, Dataset, or Prompt
>
> **Slug:** `BRAINTRUST_DELETE_RESOURCE`
>
> Delete one Braintrust experiment, dataset, or prompt by UUID for explicit lifecycle cleanup.
>
> #### Input Parameters
>
> **resource\_id** `string` *required*
>
> UUID of the selected resource to delete.
>
> **resource\_type** `string ("experiment" | "dataset" | "prompt")` *required*
>
> Kind of Braintrust resource to delete.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Fetch Events
>
> **Slug:** `BRAINTRUST_FETCH_EVENTS`
>
> Fetch one newest-to-oldest page of project log, experiment, or dataset events.
>
> #### Input Parameters
>
> **limit** `integer`
>
> Maximum full traces to request. A page can contain more events because one trace may contain multiple events.
>
> **version** `string`
>
> Optional historical transaction version snapshot to fetch.
>
> **next\_cursor** `string`
>
> Opaque continuation token returned by a previous call. Omit it for the first page and pass it back unchanged to continue newest-to-oldest.
>
> **resource\_id** `string` *required*
>
> Project, experiment, or dataset UUID matching resource\_type.
>
> **resource\_type** `string ("project_logs" | "experiment" | "dataset")` *required*
>
> Event stream to fetch.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Get Braintrust Resource
>
> **Slug:** `BRAINTRUST_GET_RESOURCE`
>
> Get one project, experiment, dataset, prompt, or function by UUID.
>
> #### Input Parameters
>
> **resource\_id** `string` *required*
>
> UUID of the selected Braintrust resource.
>
> **resource\_type** `string ("project" | "experiment" | "dataset" | "prompt" | "function")` *required*
>
> Kind of Braintrust resource to retrieve.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Insert Dataset Events
>
> **Slug:** `BRAINTRUST_INSERT_DATASET_EVENTS`
>
> Upsert a batch of examples into an existing Braintrust dataset. New or omitted IDs insert events; an existing supplied ID replaces that event's current values.
>
> #### Input Parameters
>
> **events** `array` *required*
>
> Dataset examples to upsert, in the order represented by returned row IDs. An event with an existing supplied ID replaces that event's current values.
>
> **dataset\_id** `string` *required*
>
> UUID of the Braintrust dataset that will receive the examples.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Insert Experiment Events
>
> **Slug:** `BRAINTRUST_INSERT_EXPERIMENT_EVENTS`
>
> Upsert a batch of result events into an existing Braintrust experiment. New or omitted IDs insert events; an existing supplied ID replaces that event's current values.
>
> #### Input Parameters
>
> **events** `array` *required*
>
> Experiment result events to upsert. Row IDs in the response align with this list in the same order. An event with an existing supplied ID replaces that event's current values.
>
> **experiment\_id** `string` *required*
>
> UUID of the Braintrust experiment that will receive the events.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] List Braintrust Resources
>
> **Slug:** `BRAINTRUST_LIST_RESOURCES`
>
> List and filter one page of projects, experiments, datasets, prompts, or functions visible to the connected Braintrust API key.
>
> #### Input Parameters
>
> **ids** `array`
>
> Restrict results to these resource UUIDs.
>
> **name** `string`
>
> Exact resource name to match for the selected resource type.
>
> **slug** `string`
>
> Restrict prompts or functions to this slug.
>
> **limit** `integer`
>
> Maximum number of objects to request in this page.
>
> **version** `string`
>
> Restrict prompts or functions to this version identifier.
>
> **org\_name** `string`
>
> Optional organization-name filter. It does not switch the API key's active data-plane organization.
>
> **project\_id** `string`
>
> Restrict experiments, datasets, prompts, or functions to this project UUID.
>
> **environment** `string`
>
> Restrict prompts or functions to this environment slug.
>
> **next\_cursor** `string`
>
> Opaque continuation token returned by a previous call. Omit for the first page and pass it back unchanged to continue.
>
> **project\_name** `string`
>
> Restrict experiments, datasets, prompts, or functions to this project name.
>
> **resource\_type** `string ("project" | "experiment" | "dataset" | "prompt" | "function")` *required*
>
> Kind of Braintrust resource to list.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Query Braintrust Data
>
> **Slug:** `BRAINTRUST_QUERY_DATA`
>
> Run one bounded synchronous read-only BTQL query over Braintrust project logs, an experiment, or a dataset. The first-page query must end with a positive LIMIT; continue later pages by passing the returned next\_cursor unchanged.
>
> #### Input Parameters
>
> **query** `string`
>
> Read-only BTQL SELECT query. It must query project\_logs(...), experiment(...), or dataset(...) and end with an explicit positive LIMIT. Required on the first page; omit it when passing next\_cursor.
>
> **version** `string`
>
> Optional historical data version for a reproducible query.
>
> **lint\_mode** `string ("default" | "strict")`
>
> Braintrust query linting mode.
>
> **tz\_offset** `number`
>
> Optional timezone offset in minutes using JavaScript's Date.getTimezoneOffset convention: positive values are behind UTC. For example, 480 is US Pacific Standard Time (UTC-08:00).
>
> **next\_cursor** `string`
>
> Opaque continuation token returned by a previous QUERY\_DATA call. Pass it back unchanged; the original query and options are carried authoritatively.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Summarize Experiment or Dataset
>
> **Slug:** `BRAINTRUST_SUMMARIZE_RESOURCE`
>
> Return metadata and optional aggregate summaries for one Braintrust experiment or dataset.
>
> #### Input Parameters
>
> **resource\_id** `string` *required*
>
> Experiment or dataset UUID matching resource\_type.
>
> **resource\_type** `string ("experiment" | "dataset")` *required*
>
> Resource summary to calculate.
>
> **include\_aggregates** `boolean`
>
> For experiments, summarize scores and metrics; for datasets, summarize data. Set false to return metadata only.
>
> **comparison\_experiment\_id** `string`
>
> Optional baseline experiment UUID. Valid only when resource\_type is experiment and include\_aggregates is true. When omitted, Braintrust may select the configured baseline or a recent experiment.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Update Dataset
>
> **Slug:** `BRAINTRUST_UPDATE_DATASET`
>
> Update the name, description, tags, or metadata of a Braintrust dataset. At least one non-null mutable field is required.
>
> #### Input Parameters
>
> **name** `string`
>
> New dataset name.
>
> **tags** `array`
>
> Complete replacement list of dataset tags.
>
> **metadata** `object`
>
> Metadata fields to deep-merge into the dataset's existing metadata.
>
> **dataset\_id** `string` *required*
>
> UUID of the dataset to update.
>
> **description** `string`
>
> New dataset description.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Update Experiment
>
> **Slug:** `BRAINTRUST_UPDATE_EXPERIMENT`
>
> Update selected metadata and linkage fields on a Braintrust experiment. At least one non-null mutable field is required.
>
> #### Input Parameters
>
> **name** `string`
>
> New experiment name.
>
> **tags** `array`
>
> Complete replacement list of experiment tags.
>
> **public** `boolean`
>
> Whether the experiment is publicly viewable.
>
> **metadata** `object`
>
> User-controlled metadata fields to deep-merge into existing metadata.
>
> **dataset\_id** `string`
>
> UUID of the dataset to link to the experiment.
>
> **description** `string`
>
> New experiment description.
>
> **experiment\_id** `string` *required*
>
> UUID of the experiment to update.
>
> **dataset\_version** `string`
>
> Version of the linked dataset used by the experiment.
>
> **base\_experiment\_id** `string`
>
> UUID of the default baseline experiment used for comparisons.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Update Project
>
> **Slug:** `BRAINTRUST_UPDATE_PROJECT`
>
> Update selected fields of a Braintrust project. At least one non-null mutable field is required: name, description, or settings.
>
> #### Input Parameters
>
> **name** `string`
>
> New project name.
>
> **settings** `object`
>
> Complete replacement project settings. Include every existing setting that must be retained.
>
> **project\_id** `string` *required*
>
> UUID of the project to update.
>
> **description** `string`
>
> New project description.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not

> [accordion] Update Prompt
>
> **Slug:** `BRAINTRUST_UPDATE_PROMPT`
>
> Update a Braintrust prompt's metadata or configuration without invoking it. At least one non-null mutable field is required.
>
> #### Input Parameters
>
> **name** `string`
>
> New prompt name.
>
> **slug** `string`
>
> New unique prompt slug.
>
> **tags** `array`
>
> Complete replacement list of prompt tags.
>
> **prompt\_id** `string` *required*
>
> UUID of the prompt to update.
>
> **description** `string`
>
> New prompt description.
>
> **prompt\_data** `object`
>
> Prompt block, model options, or parser fields to deep-merge. MCP attachments and tool\_functions are unsupported and rejected.
>
> **template\_format** `string ("mustache" | "nunjucks" | "none")`
>
> Replacement template syntax, stored inside the provider's prompt\_data object.
>
> #### Output
>
> **data** `string` *required*
>
> Data from the action execution
>
> **error** `string`
>
> Error if any occurred during the execution of the action
>
> **successful** `boolean` *required*
>
> Whether or not the action execution was successful or not
