[Bug]: Enforce non-empty `parts` in `Artifact` model to match A2A Specification

[punkyoon]

What happened?

The current implementation of the Artifact model in a2a/types.py allows the parts list to be empty. However, the A2A Protocol Specification (Section 4.17) explicitly states that the parts array:

"Must contain at least one part"

This lack of validation allows agents (e.g., strands-agents) to emit invalid artifacts with parts=[], which can cause issues for downstream consumers expecting valid data.

from uuid import uuid4

from a2a.types import Artifact

# This executes successfully but creates an invalid artifact according to spec
invalid_artifact = Artifact(
    artifact_id="test-id",
    parts=[],  # Empty list is currently allowed
    metadata={}
)

The Artifact model should enforce the specification constraints. Initializing an artifact with an empty parts list should raise a ValidationError.

Comparison with Other SDKs

The Java SDK already enforces this constraint. As seen in a2a-java, the validation is strictly applied:

a2a-java Artifact.java (v0.3.3.Final)

@JsonInclude(JsonInclude.Include.NON_ABSENT)
@JsonIgnoreProperties(ignoreUnknown = true)
public record Artifact(String artifactId, String name, String description, List<Part<?>> parts, Map<String, Object> metadata,
                       List<String> extensions) {

    public Artifact {
        Assert.checkNotNullParam("artifactId", artifactId);
        Assert.checkNotNullParam("parts", parts);
        if (parts.isEmpty()) {
            throw new IllegalArgumentException("Parts cannot be empty"); // !!HERE!!
        }
    }
...

The Python SDK should maintain parity with the Java SDK and the official specification.

Suggested Solution

Since the project uses Pydantic V2, we should add a min_length=1 constraint to the field definition.

as-is

https://github.com/a2aproject/a2a-python/blob/v0.3.22/src/a2a/types.py#L1372-L1400

to-be

+ from typing import Annotated
+ from pydantic import Field

class Artifact(A2ABaseModel):
    """
    Represents a file, data structure, or other resource generated by an agent during a task.
    """

    artifact_id: str
    """
    A unique identifier (e.g. UUID) for the artifact within the scope of the task.
    """
    description: str | None = None
    """
    An optional, human-readable description of the artifact.
    """
    extensions: list[str] | None = None
    """
    The URIs of extensions that are relevant to this artifact.
    """
    metadata: dict[str, Any] | None = None
    """
    Optional metadata for extensions. The key is an extension-specific identifier.
    """
    name: str | None = None
    """
    An optional, human-readable name for the artifact.
    """
-   parts: list[Part]
+   parts: Annotated[list[Part], Field(min_length=1)]
    """
    An array of content parts that make up the artifact.
    """

Relevant log output

Code of Conduct

• I agree to follow this project's Code of Conduct