Positron Assistant

Workbench | Preview

WarningPreview feature

This feature is in preview. Preview features are unsupported and may face breaking changes in a future release. Any issues found in the feature will be addressed during the regular release schedule; they will not result in immediate patches or hotfixes.

We encourage customers to try these features and we welcome any feedback via Posit Support, but we recommend that the feature not be used in production until it is in general availability (i.e., officially released as a full feature). To provide feedback, please email your Posit Customer Success representative or and specify that you are trialing this feature.

Positron Assistant is an AI coding assistant built specifically for data science workflows. Users can configure model providers like Anthropic or GitHub Copilot, or administrators can configure Workbench-managed credentials for Amazon Bedrock or Snowflake Cortex.

Settings

Enable Positron Assistant

Positron Assistant is an opt-in feature, and not enabled by default. To enable Positron Assistant for all users, specify the following in the positron-user-settings.json file:

/etc/rstudio/positron-user-settings.json
{
  "positron.assistant.enable": true
}

Disable Positron Assistant

Positron Assistant is disabled by default, but can be enabled by individual users. To enforce that Positron Assistant remains disabled, specify the following in Enforced Settings:

/etc/rstudio/enforced-settings.json
{
  "positron.assistant.enable": false
}

Manage models

Positron Assistant supports multiple language model providers, each offering various models. Administrators can customize which models are available and set default preferences.

Model preferences

Set a preferred model to have it pre-selected when users open a new chat or start a new session.

Global preference

Use the positron.assistant.models.preference.global setting to specify a model that is pre-selected across all providers. This model is used if available, unless a provider-specific preference is configured.

Matching supports partial matches on model ID or name.

/etc/rstudio/positron-user-settings.json
{
  "positron.assistant.models.preference.global": "sonnet"
}

Example values:

  • sonnet — Use any model with “sonnet” in its name
  • Claude Sonnet 4.5 — Use Claude Sonnet 4.5 if available from any provider
  • gpt-4o — Use GPT-4o from any provider that offers it
Provider-specific preferences

Use the positron.assistant.models.preference.byProvider setting to set default models for specific providers. Provider-specific preferences take precedence over the global preference.

Matching supports partial matches on model ID or name.

/etc/rstudio/positron-user-settings.json
{
  "positron.assistant.models.preference.byProvider": {
    "anthropic-api": "Claude Sonnet 4.5",
    "openai-api": "gpt-4o"
  }
}

Filtering available models

Use the positron.assistant.models.include setting to enforce strict restrictions on which models are available. When configured, only models matching at least one of the specified patterns are available to Positron Assistant.

Matching supports wildcards (*) and regular expressions for advanced filtering.

Note

This setting does not affect GitHub Copilot models.

/etc/rstudio/positron-user-settings.json
{
  "positron.assistant.models.include": [
    "claude",
    "GPT-4o",
    "^claude.*opus"
  ]
}

Example values:

  • claude — Match all Claude models
  • GPT-4o — Match this specific model
  • ^claude.*opus — Match Claude Opus models (regex)

Custom models

Use the positron.assistant.models.custom setting to manually configure models for providers. These models are used instead of retrieving the model listing from the provider.

Use provider IDs as keys, with arrays of model configurations as values.

Note

This setting does not affect GitHub Copilot models.

/etc/rstudio/positron-user-settings.json
{
  "positron.assistant.models.custom": {
    "openai-compatible": [
      {
        "name": "My Custom Model",
        "identifier": "my-custom-model",
        "maxInputTokens": 200000,
        "maxOutputTokens": 8192
      }
    ]
  }
}

Language model providers

Amazon Bedrock

To set up Amazon Bedrock with Positron Assistant:

  1. Configure AWS Credentials for Workbench, if not already configured.

  2. Configure AWS Identity and access management (IAM) for Amazon Bedrock policies to allow access to Bedrock models.

  3. Enable Positron Assistant with Amazon Bedrock as a language model provider, using the following settings:

    /etc/rstudio/positron-user-settings.json
    {
  "positron.assistant.enable": true,
  "positron.assistant.enabledProviders": ["amazon-bedrock"]
}

Snowflake Cortex

To set up Snowflake Cortex with Positron Assistant:

  1. Integrate Workbench with Snowflake

  2. Configure Snowflake access to Snowflake Cortex

  3. Enable Positron Assistant with Snowflake Cortex as a language model provider, using the following settings:

    /etc/rstudio/positron-user-settings.json
    {
  "positron.assistant.enable": true,
  "positron.assistant.enabledProviders": ["snowflake-cortex"]
}

Other providers

Refer to the Positron Assistant documentation for information on configuring additional language model providers.

Back to top