`NamedTuple` parametrized with a `ParamSpec` can't be used as an annotation

[sbrudenell]

Bug Report

NamedTuple parametrized with a ParamSpec can't be used as an annotation.

It's probably easiest to look at my repro to see what I mean.

I know that generic support for NamedTuple is new-ish as of 3.11, but I couldn't find any documentation that says ParamSpec is somehow still unsupported.

To Reproduce

from collections.abc import Callable
from typing import NamedTuple


class T[**P](NamedTuple):
    plugin: Callable[P, int]


def f[**P](a: T[P]) -> None:
    pass

I note that if I change T from a NamedTuple to a dataclass, there's no error.

Expected Behavior

$ mypy t.py                     
Success: no issues found in 1 source file

Actual Behavior

$ mypy t.py
t.py:9: error: Invalid location for ParamSpec "P"  [valid-type]
t.py:9: note: You can use ParamSpec as the first argument to Callable, e.g., "Callable[P, int]"
Found 1 error in 1 file (checked 1 source file)

The error is referring to the T[P] annotation.

Your Environment

• Mypy version used: mypy 1.17.1 (compiled: yes)
• Mypy command-line flags: mypy
• Mypy configuration options from mypy.ini (and other config files): none
• Python version used: Python 3.13.5

[sterliakov]

mypy/mypy/typeanal.py

Lines 865 to 881 in b17027e

	if tup is not None:
	# The class has a Tuple[...] base class so it will be
	# represented as a tuple type.
	if info.special_alias:
	return instantiate_type_alias(
	info.special_alias,
	# TODO: should we allow NamedTuples generic in ParamSpec?
	self.anal_array(args, allow_unpack=True),
	self.fail,
	False,
	ctx,
	self.options,
	use_standard_error=True,
	)
	return tup.copy_modified(
	items=self.anal_array(tup.items, allow_unpack=True), fallback=instance
	)

I guess the answer to this "TODO" (and for TypedDict below too) should be "yes"...