[Feat]: Feature Request: Allow custom extensions / metadata in AgentCard + richer structured response parts in Host → Agent communication

[AnkurKumarShukla]

Is your feature request related to a problem? Please describe.

In real multi-agent orchestration (Host agent calling other A2A agents), we need to expose additional machine-readable info in the AgentCard (e.g., payment / premium access terms like scheme, receiver wallet, amount, network, facilitator URL).

Currently the AgentCard schema is strict and does not allow custom metadata or extensions at the AgentCard level, so any attempt to attach extra fields fails validation or breaks interoperability between clients/hosts.

Because of this limitation:

• premium agents cannot advertise payment terms in a standards-compliant way
• host agents cannot reliably discover whether an agent is free vs paid
• developers are forced to hack data into unsupported fields (causing schema validation errors)

Describe the solution you'd like

Add an official, schema-valid extension mechanism for AgentCard so agents can publish additional machine-readable information.

Preferred solution:

1. Allow AgentCard.extensions: list[AgentExtension] (or similar) at AgentCard level, not only for Message/Artifacts.
2. Or allow AgentCard.metadata: dict[str, Any] for vendor-defined metadata.

This would enable use cases like:

• premium/paid agents advertising payment terms (x402-style)
• marketplace discovery (pricing, network, scheme)
• custom capabilities that are not skills

Also, document a recommended pattern for agents to return structured outputs (DataPart) so host agents can interpret responses programmatically in orchestration workflows.

Describe alternatives you've considered

No response

Additional context

No response

Code of Conduct

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