You’re very close — the

invoke(["source", "freshness"])

syntax is correct. The error you’re seeing is coming from dbt, not Prefect: dbt can’t find a

dbt_project.yml

in the working directory. That’s why

build

may work for you (depending on how your runner is initialized), but

source freshness

fails. Ways to fix it: - Run from your dbt project directory (or set

DBT_PROJECT_DIR

) - Pass the project (and profiles) directories as flags in your invoke call - Configure the runner with

PrefectDbtSettings

that specify the project/profiles dirs Examples: 1) Pass flags inline

runner = PrefectDbtRunner(...)
runner.invoke([
    "source", "freshness",
    "--project-dir", "/path/to/your/dbt_project",
    "--profiles-dir", "/path/to/your/.dbt"
])

2) Configure once via settings

from prefect_dbt import PrefectDbtRunner, PrefectDbtSettings

settings = PrefectDbtSettings(
    project_dir="/path/to/your/dbt_project",
    profiles_dir="/path/to/your/.dbt",
)
runner = PrefectDbtRunner(settings=settings)
runner.invoke(["source", "freshness"])

Notes: - The

'sourcefreshness'

you see in the error string is just how the error message is formatted — dbt still expects two tokens (

source

and

freshness

) and your call is correct. - You can also set

DBT_PROJECT_DIR

and/or

DBT_PROFILES_DIR

in your environment if you prefer not to pass flags. If you’d rather use the helper task, Prefect also exposes:

from prefect_dbt.cli.commands import run_dbt_source_freshness

await run_dbt_source_freshness(
    project_dir="/path/to/your/dbt_project",
    profiles_dir="/path/to/your/.dbt"
)

Docs for the Prefect dbt integration: Prefect dbt integration docs If this still fails, can you share: - How you’re constructing

PrefectDbtRunner

(are you passing a

manifest

?) - Where

dbt_project.yml

lives at runtime (container path, working dir, etc.) - Whether you have

DBT_PROJECT_DIR

or

DBT_PROFILES_DIR

set in the environment?

Thanks for the details — your syntax is correct. The

'sourcefreshness'

you see is just how the error message is concatenated; dbt still expects two tokens (

"source"

,

"freshness"

). Given you’re already passing

PrefectDbtSettings

, this looks like the `project_dir`/`profiles_dir` from settings aren’t being picked up for this specific subcommand. Two reliable workarounds: - Pass the dirs as flags in the invoke call (putting global flags before the subcommand tends to be most robust):

runner.invoke([
    "--project-dir", "/some/path",
    "--profiles-dir", "/some/path",
    "source", "freshness",
])

If that doesn’t work in your environment, try flags after the subcommand:

runner.invoke([
    "source", "freshness",
    "--project-dir", "/some/path",
    "--profiles-dir", "/some/path",
])

- Use the helper task which wires this up for you:

from prefect_dbt.cli.commands import run_dbt_source_freshness

await run_dbt_source_freshness(
    project_dir="/some/path",
    profiles_dir="/some/path",
)

If you still see the same error: - What versions of

prefect-dbt

and

dbt-core

are you running? - Can you confirm that

/some/path/dbt_project.yml

exists inside the runtime environment (e.g., container path)? - Please try

print(runner.settings)

right before invoke to confirm the values at runtime. Docs for reference: Prefect dbt integration docs