BrowserStack AI Evals
Settings & Configuration

Project Settings

Configure general project properties, AI features, and project lifecycle.

Project Settings

Access via Settings → General in your project.

Project Name

Rename your project at any time. The project name appears in the dashboard, API responses, and SDK trace metadata.

  1. Go to Settings → General
  2. Click the project name field and enter the new name
  3. Click Save

Project names must be unique within your organization.

Project Hostname / Project ID

Each project has a stable project ID (slug) used in API paths and SDK configuration. The hostname is read-only and is derived when the project is created. You can copy it from the General settings page.

Use the project ID when:

  • Configuring the SDK: publicKey and secretKey are scoped to a project
  • Building API requests: /api/public/projects/{projectId}/...
  • Setting up CI/CD pipelines with environment-specific projects

Team Assignment

Assign a project to a team within your organization. Teams are groups of members managed at the organization level. Assigning a project to a team:

  • Gives all team members access to the project at their team-level role
  • Allows team-level API keys to be used for this project
  • Appears in team-scoped dashboards and reporting

Change the team assignment in Settings → General → Team.

AI Features Toggle

The AI Features toggle enables or disables LLM-powered features at the project level, including:

  • LLM-as-judge evaluators
  • Playground (LLM prompt testing)
  • AI-assisted annotation in human review queues

Disabling this toggle prevents all AI feature usage for the project, regardless of individual LLM connections configured. Re-enabling it restores access immediately.

AI features require at least one LLM Connection to be configured and a default evaluation model to be set.

Score Labels / Evaluation Config

Under Settings → Labels / Evaluation, configure the score dimension schema for your project:

  • Score name — the metric name (e.g., faithfulness, toxicity)
  • Data typeNUMERIC (0–1 range), BOOLEAN, or CATEGORICAL
  • Categories — for CATEGORICAL types, define the allowed values
  • Description — optional description for annotators

Score configs are used by human review queues and evaluators to validate scores.

Protected Prompt Labels

Under Settings → Protected Prompt Labels (if enabled for your plan), you can lock specific prompt labels so they can only be modified by project Owners and Admins. For example, mark production as protected to prevent accidental label changes.

Batch Exports

Under Settings → Exports, view and download all batch export jobs for your project. Supported formats include CSV and JSON.

Danger Zone — Delete Project

Deleting a project is permanent and cannot be undone. All traces, scores, datasets, experiments, prompts, and configuration will be destroyed.

To delete a project:

  1. Go to Settings → General
  2. Scroll to the Danger Zone section
  3. Click Delete this project
  4. Confirm the project name when prompted

Only project Owners can delete a project.