Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.
Python agentagentic-aiagentic-frameworkagentic-workflowaiautomationclaudedeepseekgenaigptllmlow-codemcpnextjsno-codeopenaipythonskillsworkflow
import pytest
from pydantic import ValidationError
from sqlalchemy.dialects import sqlite
from models.human_input_v2 import IMIdentityRawPayload
from models.types import FrozenPydanticModelColumn
def test_im_identity_raw_payload_accepts_only_a_strict_json_object_and_is_frozen() -> None:
raw_payload = IMIdentityRawPayload({"provider_user": {"id": "provider-user-1"}})
assert raw_payload.root == {"provider_user": {"id": "provider-user-1"}}
assert raw_payload.model_config["frozen"] is True
assert raw_payload.model_config["strict"] is True
assert raw_payload.model_config["validate_default"] is True
with pytest.raises(ValidationError):
IMIdentityRawPayload.model_validate([("provider_user", {"id": "provider-user-1"})])
with pytest.raises(ValidationError):
raw_payload.root = {}
def test_im_identity_raw_payload_round_trips_through_its_persistence_column() -> None:
raw_payload = IMIdentityRawPayload(
{
"provider_user": {"id": "provider-user-1"},
"roles": ["reviewer"],
}
)
column_type = FrozenPydanticModelColumn(IMIdentityRawPayload)
serialized = column_type.process_bind_param(raw_payload, sqlite.dialect())
restored = column_type.process_result_value(serialized, sqlite.dialect())
assert restored == raw_payload