solidworks_mcp.ui.routers.local_model¶

solidworks_mcp.ui.routers.local_model ¶

Local Ollama model probe / pull / query routes for the Prefab CAD dashboard.

Attributes¶

router `module-attribute` ¶

router = APIRouter()

Classes¶

LocalAgentResult ¶

Bases: BaseModel, Generic[_T]

Typed envelope wrapping a structured pydantic-ai agent response.

data holds the validated result_type instance; config echoes back the LocalLLMConfig used so callers can log or audit provenance. Set success=False and error when the agent returned a RecoverableFailure or raised an exception.

Attributes:

Name	Type	Description
`config`	`LocalLLMConfig`	The config value.
`data`	`Any`	The data value.
`error`	`str \| None`	The error value.
`retry_hint`	`str \| None`	The retry hint value.
`success`	`bool`	The success value.

LocalModelProbeResult ¶

Bases: BaseModel

Full hardware-detection and Ollama availability result.

Returned by probe_local_model() and serialised as the JSON response from GET /api/ui/local-model/probe. The to_config() helper converts directly into a LocalLLMConfig ready for run_local_agent().

Attributes:

Name	Type	Description
`all_tiers`	`dict[str, str]`	The all tiers value.
`available`	`bool`	The available value.
`endpoint`	`str`	The endpoint value.
`label`	`str`	The label value.
`ollama_model`	`str`	The ollama model value.
`openai_endpoint`	`str`	The openai endpoint value.
`pull_command`	`str`	The pull command value.
`pulled_models`	`list[str]`	The pulled models value.
`ram_gb`	`float`	The ram gb value.
`service_model`	`str`	The service model value.
`status_message`	`str`	The status message value.
`tier`	`Literal['small', 'balanced', 'large']`	The tier value.
`tier_already_pulled`	`bool`	The tier already pulled value.
`vram_gb`	`float`	The vram gb value.

Functions¶

to_config ¶

to_config() -> LocalLLMConfig

Convert probe result into a ready-to-use LocalLLMConfig.

Returns:

Name	Type	Description
`LocalLLMConfig`	`LocalLLMConfig`	The result produced by the operation.

Source code in src/solidworks_mcp/ui/local_llm.py

def to_config(self) -> LocalLLMConfig:
    """Convert probe result into a ready-to-use ``LocalLLMConfig``.

    Returns:
        LocalLLMConfig: The result produced by the operation.
    """
    return LocalLLMConfig(
        endpoint=self.endpoint,
        openai_endpoint=self.openai_endpoint,
        tier=self.tier,
        ollama_model=self.ollama_model,
        service_model=self.service_model,
    )

LocalModelPullRequest ¶

Bases: BaseModel

Request body for POST /api/ui/local-model/pull.

Attributes:

Name	Type	Description
`endpoint`	`str \| None`	The endpoint value.
`model`	`str`	The model value.

LocalModelPullResult ¶

Bases: BaseModel

Result from POST /api/ui/local-model/pull.

Attributes:

Name	Type	Description
`error`	`str \| None`	The error value.
`model`	`str`	The model value.
`queued`	`bool`	The queued value.
`response`	`dict[str, Any] \| None`	The response value.

LocalModelQueryRequest ¶

Bases: BaseModel

Request body for POST /api/ui/local-model/query.

Attributes:

Name	Type	Description
`endpoint`	`str \| None`	The endpoint value.
`model`	`str \| None`	The model value.
`prompt`	`str`	The prompt value.
`system_prompt`	`str`	The system prompt value.

Functions¶

probe_local_model_endpoint `async` ¶

probe_local_model_endpoint() -> LocalModelProbeResult

Probe for a running Ollama server and recommend the best Gemma model tier.

Source code in src/solidworks_mcp/ui/routers/local_model.py

@router.get("/api/ui/local-model/probe")
async def probe_local_model_endpoint() -> LocalModelProbeResult:
    """Probe for a running Ollama server and recommend the best Gemma model tier."""
    from ..local_llm import probe_local_model  # noqa: PLC0415

    return await probe_local_model()

pull_local_model_endpoint `async` ¶

pull_local_model_endpoint(payload: LocalModelPullRequest) -> LocalModelPullResult

Trigger an Ollama pull for the specified model name.

Source code in src/solidworks_mcp/ui/routers/local_model.py

@router.post("/api/ui/local-model/pull")
async def pull_local_model_endpoint(
    payload: LocalModelPullRequest,
) -> LocalModelPullResult:
    """Trigger an Ollama pull for the specified model name."""
    from ..local_llm import pull_ollama_model  # noqa: PLC0415

    return await pull_ollama_model(model=payload.model, endpoint=payload.endpoint)

query_local_model_endpoint `async` ¶

query_local_model_endpoint(payload: LocalModelQueryRequest) -> LocalAgentResult

Run a free-form prompt against the local Ollama model.

Source code in src/solidworks_mcp/ui/routers/local_model.py

@router.post("/api/ui/local-model/query")
async def query_local_model_endpoint(
    payload: LocalModelQueryRequest,
) -> LocalAgentResult:
    """Run a free-form prompt against the local Ollama model."""
    from pydantic import BaseModel as _BaseModel

    from ..local_llm import LocalLLMConfig, run_local_agent  # noqa: PLC0415

    class _FreeFormResponse(_BaseModel):
        text: str

    config = LocalLLMConfig.from_env()
    if payload.endpoint:
        config = config.model_copy(
            update={
                "endpoint": payload.endpoint,
                "openai_endpoint": f"{payload.endpoint}/v1",
            }
        )
    if payload.model:
        service_model = (
            payload.model
            if payload.model.startswith("local:")
            else f"local:{payload.model}"
        )
        config = config.model_copy(update={"service_model": service_model})

    return await run_local_agent(
        system_prompt=payload.system_prompt,
        user_prompt=payload.prompt,
        result_type=_FreeFormResponse,
        config=config,
    )

solidworks_mcp.ui.routers.local_model¶