Configuration

Read and update API Performance evaluation configuration from the Sapient CLI.

Configuration controls which models, scheduled targets, endpoints, custom evals, environment profiles, and API base URL Sapient uses for API Performance.

Commands

Command	Description
sapient api-performance config retrieve	Get the API Performance config.
sapient api-performance config update	Update the API Performance config.

Use Targets to discover valid target_type, execution_platform, and model_id combinations before updating config.

Retrieve config

sapient api-performance config retrieve

Retrieve config for one source:

sapient api-performance config retrieve --integration-id int_123

Example response:

{
  "models": ["claude-sonnet-4.5"],
  "targets": [],
  "endpoint_ids": ["endpoint_123"],
  "custom_eval_ids": ["custom_eval_123"],
  "env_vars": {},
  "env_profiles": [
    {
      "id": "prod",
      "name": "Production",
      "env_var_keys": ["API_KEY"]
    }
  ],
  "default_env_profile_id": "prod",
  "api_base_url": "https://api.example.com",
  "can_configure_per_model_schedules": true
}

Update models and eval selection

sapient api-performance config update \
  --models '["claude-sonnet-4.5"]' \
  --endpoint-ids '["endpoint_123"]' \
  --custom-eval-ids '["custom_eval_123"]'

Update custom eval selection:

sapient api-performance config update \
  --custom-eval-ids '["custom_eval_123","custom_eval_456"]'

Update scheduled targets

Use --targets for scheduled text or agent targets:

sapient api-performance config update \
  --targets '[{"target_type":"agent","execution_platform":"claude-code","model_id":"global.anthropic.claude-sonnet-4-6","frequency":"daily","enabled":true}]'

Target fields include target_type, execution_platform, model_id, frequency, enabled, next_run_at, and integration_id.

For text model evals, use execution_platform set to openrouter and a text model ID from the target catalog.

Set base URL and environment variables

sapient api-performance config update \
  --api-base-url "https://api.example.com" \
  --env-vars '{"API_KEY":"sk_test_123","API_BASE_URL":"https://api.example.com"}'

Read responses include environment variable keys, not secret values. Update requests can send environment variables, but Sapient does not return stored secret values.

Configure environment profiles

Environment profiles let you store named sets of environment variables and choose which profiles a specific eval can use.

sapient api-performance config update \
  --env-profiles '[{"id":"prod","name":"Production","env_vars":{"API_KEY":"sk_live_123"}},{"id":"sandbox","name":"Sandbox","env_vars":{"API_KEY":"sk_test_123"}}]' \
  --default-env-profile-id prod

Read responses expose each profile’s id, name, and env_var_keys. Secret values are not returned.

Fields

Flag	Description
--models	Model IDs to include as a JSON array.
--targets	Scheduled target objects as a JSON array.
--endpoint-ids	Endpoint or operation eval IDs to include as a JSON array.
--custom-eval-ids	Custom eval IDs to include as a JSON array.
--env-vars	Environment variable map as JSON.
--env-profiles	Environment profile objects as a JSON array.
--default-env-profile-id	Default environment profile ID.
--integration-id	Source ID for source-specific config.
--api-base-url	Base URL available to eval runs.