StreamFlow Mypy Sub-Skill: var-annotated

Error: [var-annotated] — mypy cannot infer a variable's type. Fix: add an explicit annotation (variable: Type = value). If the correct type requires forbidden types, skip the error.

Workflow

1. Read surrounding code — what gets stored/retrieved from the variable?
2. Determine the concrete type from usage, signatures, or similar patterns nearby
3. Check: does the type need Any or other forbidden types? If YES → skip
4. Apply: variable: ConcreteType = value
5. Validate & commit: Load the StreamFlow Mypy Type Checking skill — General Workflow steps 3–5

Fix Patterns

Empty collections — examine what gets stored:

cache: dict[str, Port] = {}
items: list[Token] = []
ports: dict[str, list[str]] = {}

Generic types — specify type parameters:

future: asyncio.Future[str] = asyncio.Future()
cache: LRUCache[int, int] = LRUCache(maxsize=5)

For generic type parameters, prefer covariant alternatives when you only need read access: use Sequence[T] over MutableSequence[T], Mapping[K, V] over MutableMapping[K, V]. Check typeshed for variance of standard library types.

StreamFlow objects — import the concrete type:

from streamflow.core.workflow import Port, Token
from streamflow.core.deployment import ExecutionLocation

storage: dict[str, Token] = {}

Common StreamFlow types: Port, Token (streamflow.core.workflow) · Storage (streamflow.data.utils) · ExecutionLocation (streamflow.core.deployment)

Nested dicts with known keys — prefer TypedDict:

from typing import TypedDict

class Config(TypedDict):
    name: str
    count: int

config: Config = {"name": "test", "count": 42}

# Only use generic dict when keys are truly dynamic:
extensions: dict[str, dict[str, str]] = {}

Don't guess types — always verify by reading code. Don't use bare generics (list instead of list[str]).

See Also

• StreamFlow Mypy Type Checking skill — General workflow, allowed types reference
• AGENTS.md — Forbidden types list, commit approval rule