feat: add Requesty as an OpenAI-compatible LLM provider

[Thibaultjaigu]

Add Requesty as an OpenAI-compatible LLM provider

This adds a native RequestyLLMProvider that mirrors the existing OpenRouterLLMProvider as closely as possible. Requesty is an OpenAI-compatible LLM gateway, so the implementation reuses the exact same chat-completions logic.

New provider

packages/server/src/llm/requesty.ts — export class RequestyLLMProvider implements LLMProvider:

• readonly name = 'requesty'
• apiKey from opts.apiKey || process.env.REQUESTY_API_KEY
• baseUrl default https://router.requesty.ai/v1
• default model openai/gpt-4o-mini (Requesty uses provider/model naming, same as OpenRouter)
• identical complete() logic to OpenRouter: OpenAI-compatible POST /chat/completions, Authorization: Bearer <key>, same messages/max_tokens/temperature handling, same timeout (AbortSignal.timeout) and error handling

Wiring sites

• packages/server/src/llm/requesty.ts — new provider class
• packages/server/src/llm/cascade.ts — import { RequestyLLMProvider } + case 'requesty': return new RequestyLLMProvider(config); in createLLMProvider
• packages/server/src/utils/config.ts — added 'requesty' to the LLMProviderSchema provider zod enum
• packages/dashboard/src/pages/Settings/types.ts — added a requesty preset to LLM_PROVIDERS (label, base URL, model list, REQUESTY_API_KEY env key)
• README.md — added Requesty to the supported providers table

Verification

• Typecheck passes: tsc --noEmit (server lint script) and tsc -b (dashboard) both exit clean
• Live smoke test: a real chat completion against https://router.requesty.ai/v1/chat/completions with model openai/gpt-4o-mini returned HTTP 200 with a valid completion

Links

• https://requesty.ai
• https://docs.requesty.ai

I work at Requesty. This mirrors the existing OpenRouter provider as closely as possible. Happy to adjust or close it if it's not a fit.