`score evaluate` with OpenAI fails with incorrect regional hostname error

[BryceLohr]

What happened?

I tried to use the score evaluate command with OpenAI, and got the following error:

$ skill-validator score evaluate --provider openai .claude/skills/back-end-development/
  scoring: back-end-development/SKILL.md
Error: scoring SKILL.md: scoring SKILL.md: API returned status 401: {
  "error": {
    "message": "Attempted to access resource with incorrect regional hostname. Please make your request to us.api.openai.com",
    "type": "invalid_request_error",
    "code": "incorrect_hostname",
    "param": null
  },
  "status": 401
}

What did you expect?

The LLM-based scoring to run on the given skill.

How to reproduce

Simply run the above command. I set the OPENAI_API_KEY, OPENAI_PROJECT_ID, and OPENAI_ORGANIZATION_ID environment variables to valid values prior to running that command. I verified that the keys were working in other code.

skill-validator score evaluate --provider openai <path>

Environment

• skill-validator version: v1.5.4
• Go version: none
• OS: Debian 13.3 in a container

Additional context

I downloaded the skill-validator binary directly from the Github releases page, and put it in /usr/local/bin inside the Linux container I'm using.

[dacharyc]

Thanks for filing the issue! Sorry about that; in my own testing, I wasn't using either of those env vars so I missed this.

I've implemented support for these env vars in the linked PR:

• OPENAI_ORG_ID - sent as a OpenAI-Organization header
• OPENAI_PROJECT_ID - sent as OpenAI-Project header

The naming varies slightly compared to what you're using here; I aligned with the OpenAI Python SDK env var naming.

I've just cut a release, v1.5.5, which includes the env var support. Let me know if this resolves your issue!