Add backward-compatible OpenAI Responses API support with opt-in env var

[Copilot]

PR #277 proposed switching the default OpenAI endpoint from /chat/completions to /responses, but the change was incomplete: the Responses API has a different request format (input vs messages) and response shape (output[0].content[0].text vs choices[0].message.content), making the PR as-is a silent breaking change.

Changes

• New export createOpenAIResponsesLanguageModel — mirrors createOpenAILanguageModel but targets the Responses API with the correct request/response handling:

  • Sends input (not messages) in the request body
  • Parses output[0].content[0].text from the response
  • Defaults to https://api.openai.com/v1/responses

• createOpenAILanguageModel unchanged — still defaults to /chat/completions; no breakage for existing users or Azure OpenAI (which may not support /responses)

• createLanguageModel opt-in flag — set OPENAI_USE_RESPONSES_API=true to route through the Responses API path; omitting it preserves the existing behavior

// Existing code — unchanged, backward compatible
const model = createOpenAILanguageModel(apiKey, "gpt-4");

// New: explicit Responses API
const model = createOpenAIResponsesLanguageModel(apiKey, "gpt-4");

// New: via env var in createLanguageModel
// OPENAI_USE_RESPONSES_API=true  →  uses /v1/responses
// (default)                      →  uses /v1/chat/completions
const model = createLanguageModel(process.env);

• Tests — 16 unit tests added (typescript/tests/model.test.mjs) using Node's built-in node:test with mocked fetch, covering both API paths and createLanguageModel routing; "test" script added to package.json

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.