Skip to content

provider-neutral A2Agent conformance coverage for openaicompat #369

Description

@A2agent-ai

Hi Fantasy maintainers — I represent A2Agent, an OpenAI- and Anthropic-compatible API gateway.

Fantasy already directs OpenAI-compatible services to the generic openaicompat layer, and your response on provider proposals indicates that provider-specific code is unnecessary when that layer works. We would like to respect that boundary and verify A2Agent as an external conformance target.

If useful, we can contribute only the scope you prefer:

  • a small provider-neutral example showing custom endpoint and key configuration;
  • recorded/mock regression cases derived from live tests for streaming, tools, reasoning and media/tool-result messages; and
  • notes on base-URL joining, model selection and sanitized errors.

We are not proposing a dedicated A2Agent provider package unless the tests demonstrate a real protocol deviation. Would a generic example or compatibility fixture be welcome, and where should it live?

Separately, if the technical evidence is accepted, would you consider a factual A2Agent link under Examples, Tested Services, Integrations, or another location you choose? Please confirm the wording/location and whether a logo is acceptable. This is optional, separate from technical review, and does not imply endorsement.

A2Agent can provide a private test key, technical/documentation support, and $20–$100 in non-cash API testing credits, with possible continuing credits for maintained tests. This support is not consideration for merging code, endorsement, referrals, or displaying a link.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions