Add ai:proxyurl to Wave AI mode config and wire proxy support through Gemini/OpenAI Chat

[Copilot]

Wave AI mode config (waveai entries) did not expose a per-mode proxy URL, so AIOptsType.ProxyURL was never populated from mode settings. As a result, proxy routing worked in some backends only when set elsewhere, and uctypes.AIModeConfig remained as an obsolete, confusing duplicate type.

• Config model + schema

  • Added ProxyURL string \json:"ai:proxyurl,omitempty"`towconfig.AIModeConfigType`.
  • Regenerated schema/waveai.json so ai:proxyurl is accepted in Wave AI mode config.

• Runtime option plumbing

  • Updated getWaveAISettings to propagate mode-level ProxyURL into uctypes.AIOptsType.ProxyURL.

• Backend proxy handling parity

  • Added proxy-aware HTTP client construction in:
    • pkg/aiusechat/gemini/gemini-backend.go
    • pkg/aiusechat/openaichat/openaichat-backend.go
  • Behavior now matches existing OpenAI/Anthropic handling: parse proxy URL, fail fast on invalid URLs, and set http.Transport.Proxy.

• Type cleanup

  • Removed unused uctypes.AIModeConfig and its helper method to eliminate stale/duplicate AI mode config definitions.

• Focused coverage

  • Added minimal tests for proxy client creation in Gemini/OpenAI Chat backends.
  • Added mode-default test ensuring ProxyURL survives provider-default resolution.

Example Wave AI mode config:

{
  "waveai": {
    "my-openai-mode": {
      "display:name": "OpenAI via Proxy",
      "ai:provider": "openai",
      "ai:model": "gpt-5-mini",
      "ai:proxyurl": "http://localhost:8080"
    }
  }
}

---

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

[cloudflare-workers-and-pages]

Deploying waveterm with    Cloudflare Pages

Latest commit:	1019c2a
Status:	✅  Deploy successful!
Preview URL:	https://912998a2.waveterm.pages.dev
Branch Preview URL:	https://copilot-add-proxyurl-to-aimo.waveterm.pages.dev

View logs