Sources

langsmith

Read transcripts from LangSmith traces.

Each LangSmith root run (trace) becomes one Scout transcript. Child runs (LLM calls, tools) become events within the transcript.

Data sources:

project - Import traces from a project (default)
dataset - Import examples from a dataset

Source

async def langsmith(
    project: str | None = None,
    dataset: str | None = None,
    from_time: datetime | None = None,
    to_time: datetime | None = None,
    tags: list[str] | None = None,
    filter: str | None = None,
    trace_filter: str | None = None,
    run_type: str | None = None,
    limit: int | None = None,
    api_key: str | None = None,
    api_url: str | None = None,
) -> AsyncIterator[Transcript]

project str | None: LangSmith project name. Optional for dataset, used as default data source otherwise.
dataset str | None: LangSmith dataset name or ID. Fetches from curated evaluation datasets instead of project traces.
from_time datetime | None: Only fetch traces created on or after this time
to_time datetime | None: Only fetch traces created before this time
tags list[str] | None: Filter by tags (all must match)
filter str | None: LangSmith filter string for run filtering. Can use filters saved from the LangSmith UI.
trace_filter str | None: Filter on root run attributes
run_type str | None: Filter by run type (llm, chain, tool, etc.)
limit int | None: Maximum number of transcripts to fetch
api_key str | None: LangSmith API key (or LANGSMITH_API_KEY env var)
api_url str | None: LangSmith API URL (or LANGSMITH_ENDPOINT env var)

logfire

Read transcripts from Logfire traces.

Each Logfire trace (collection of spans with same trace_id) becomes one Scout transcript. Child spans (LLM calls, tools) become events within the transcript.

Source

async def logfire(
    project: str | None = None,
    from_time: datetime | None = None,
    to_time: datetime | None = None,
    filter: str | None = None,
    trace_id: str | None = None,
    limit: int | None = None,
    read_token: str | None = None,
) -> AsyncIterator[Transcript]

project str | None: Logfire project name in format “org/project”. If not provided, queries across all accessible projects.
from_time datetime | None: Only fetch traces created on or after this time
to_time datetime | None: Only fetch traces created before this time
filter str | None: SQL WHERE fragment for additional filtering (e.g., “attributes->>‘gen_ai.request.model’ = ‘gpt-4o’”)
trace_id str | None: Fetch a specific trace by ID instead of querying
limit int | None: Maximum number of transcripts to fetch
read_token str | None: Logfire read token (or LOGFIRE_READ_TOKEN env var). Generate from Logfire dashboard Settings > Read Tokens.