Scout Import

Import MLflow traces into Inspect Scout transcript databases for safety scanner analysis.

Requires the optional scout extra:

pip install 'inspect-mlflow[scout]'

Usage

from inspect_mlflow.scout import import_mlflow_traces
from inspect_scout import transcripts_db

async with transcripts_db("./safety-analysis") as db:
    await db.insert(import_mlflow_traces(
        experiment_name="my-evals",
        tracking_uri="http://localhost:5000",
    ))

Data Mapping

MLflow Span

Scout Event

LLM span

ModelEvent (model, tokens, timing)

TOOL span

ToolEvent (function, arguments, result)

EVALUATOR span

ScoreEvent (value, explanation, target)

API Reference

async inspect_mlflow.scout.import_mlflow_traces(experiment_name: str | None = None, tracking_uri: str | None = None, limit: int | None = None) → AsyncIterator[Transcript]

Import MLflow traces as Scout transcripts.

Parameters:

  • experiment_name – MLflow experiment name to import from. Defaults to MLFLOW_EXPERIMENT_NAME env var or “inspect_ai”.

  • tracking_uri – MLflow tracking server URI. Defaults to MLFLOW_TRACKING_URI env var.

  • limit – Maximum number of traces to import. None for all.

Yields:

Transcript objects ready for Scout database insertion.