feat(openai-auth): add OpenAI Max device-code authentication and unified prompt caching helpers

readme.hints.md

@@ -10,6 +10,7 @@ The package is a **provider registry** built on the Vercel AI SDK (`ai` v6). The
 - Providers: anthropic, openai, google, groq, mistral, xai, perplexity, ollama
 - Anthropic prompt caching via `wrapLanguageModel` middleware (enabled by default)
 - Custom Ollama provider implementing `LanguageModelV3` directly (for think, num_ctx support)
+- OpenAI ChatGPT/Max device-code auth in `smartai.auth.openai.ts`; `openAiMaxAuth` routes OpenAI models to the ChatGPT Codex backend
 
 ### Subpath Exports
 - `@push.rocks/smartai/vision` — `analyzeImage()` using `generateText` with image content
@@ -43,6 +44,7 @@ The package is a **provider registry** built on the Vercel AI SDK (`ai` v6). The
 - `qenv.getEnvVarOnDemand()` returns a Promise — must be awaited in tests
 - OpenAI reasoning options belong in AI SDK `providerOptions`, not model construction options
 - SmartAI accepts OpenAI model IDs as plain strings, including `gpt-5.5`
+- ChatGPT-authenticated OpenAI model calls use `https://chatgpt.com/backend-api/codex` plus bearer access token and `ChatGPT-Account-ID`, not `https://api.openai.com/v1`
 
 ## Testing