data.ai-provider.get

Fetch one AIProviderConfig by UUID, scoped to the calling organization

Fetch a single AIProviderConfig by UUID, org-scoped.

Read-only — no DB mutations. Uses context.metadata["organization"] for the internal organization_id lookup; the public input is organization_uuid only.

Output shapes:

• Found: found=True + all record fields
• Not found: found=False (no other fields present; cross-org isolation upheld)

Overview

Property	Value
Workflow type	Atomic
Library	App-connection
Version	1.0

Input Schema

Field	Type	Required	Default	Description
organization_uuid	uuid	Yes	—	UUID of the organization the request is scoped to
ai_provider_config_uuid	uuid	Yes	—	UUID of the AIProviderConfig row to fetch

Output Schema

Field	Type	Required	Default	Description
found	boolean	Yes	—	True when the record exists and belongs to the org
ai_provider_config_uuid	uuid	No	—	UUID of the fetched AIProviderConfig row (present only when found=True)
connection_uuid	uuid	No	—	UUID of the linked CloudConnection (present only when found=True)
default_model	string	No	—	Default model ID configured for this provider (e.g. 'gpt-4o-mini')
models	list	No	—	User-supplied list of model IDs (e.g. for OpenRouter); None when unset
model_config	dict	No	—	Model configuration dict: temperature, max_tokens, etc. None is normalized to {}
rate_limit_rpm	integer	No	—	Requests per minute rate limit; None means unlimited
rate_limit_tpm	integer	No	—	Tokens per minute rate limit; None means unlimited
is_default	boolean	No	—	Whether this is the org's default AI provider config
enabled	boolean	No	—	Whether this provider config is active
total_requests	integer	No	—	Total number of AI requests made through this config
total_tokens	integer	No	—	Total tokens consumed (input + output)
total_cost_usd	float	No	—	Total cost in USD; converted from Numeric for JSON safety
last_used_at	string	No	—	ISO-8601 timestamp of last usage; None when never used
failure_reason	string	No	—	Engine-stamped: human-readable failure reason
failure_type	string	No	—	Engine-stamped: failure category
failed_action	string	No	—	Engine-stamped: action method that raised
failed_at_state	string	No	—	Engine-stamped: state name when the failure occurred
failed_step	string	No	—	Engine-stamped: name of the step that failed (DAG workflows)
failed_layer	integer	No	—	Engine-stamped: layer index that failed (DAG workflows)
error	string	No	—	Engine-stamped: error message on the failure envelope
error_type	string	No	—	Engine-stamped: exception class name on the failure envelope

States

State	Initial	Terminal	Success	Auto-advance	Description
pending	Yes	No	—	complete	Fetch AI provider config by UUID
completed	No	Yes	Yes	—	AI provider config fetched (or confirmed absent)
failed	No	Yes	No	—	Fetch failed

State Diagram

Transitions

From	Action	To	Description
pending	complete	completed	—
pending	fail	failed	—

API Usage

POST /api/workflows/start
Content-Type: application/json

{
  "workflow_type": "data.ai-provider.get",
  "initial_data": {
    "organization_uuid": "value",
    "ai_provider_config_uuid": "value"
  }
}