feat: add Requesty as an OpenAI-compatible LLM provider

[Thibaultjaigu]

Summary

Adds Requesty as an OpenAI-compatible LLM provider, mirroring the existing OpenRouter provider as closely as possible.

Requesty (https://requesty.ai) is an OpenAI-compatible LLM gateway. It exposes the standard /v1/chat/completions endpoint, uses Authorization: Bearer <key> auth, and provider/model naming (e.g. openai/gpt-4o-mini), so it slots into the existing OpenAI-compatible provider plumbing with no special-casing.

• OpenAI-compatible base URL: https://router.requesty.ai/v1
• Chat completions: https://router.requesty.ai/v1/chat/completions

Wiring sites (each mirrors the OpenRouter entry exactly)

Frontend (frontend/magic-web):

• packages/magic-admin/src/const/aiModel.ts — added Requesty to the ServiceProvider enum and https://router.requesty.ai/v1/chat/completions to the ServiceProviderUrl base-URL map.
• packages/magic-admin/src/pages/PlatformPackage/components/ServiceIcon/index.tsx — added Requesty to the service-icon map (reuses the existing default icon; no new binary asset).
• src/pages/superMagic/.../AddModel/providerFieldConfigs.ts — added requesty to the provider-code map, the URL-placeholder override map (https://router.requesty.ai/v1), and the API-key-placeholder map.
• src/pages/initialization/components/Step2Provider.tsx — added the Requesty provider config (defaultUrl: https://router.requesty.ai/v1) and the dropdown entry.
• src/assets/locales/en_US/initialization.json and zh_CN/initialization.json — added the Requesty label.

Backend (backend/magic-service):

• app/Domain/Provider/Entity/ValueObject/ProviderCode.php — added the Requesty enum case. It falls through the existing default arms (implementation → OpenAIModel, sort order → 999), i.e. it is handled as a generic OpenAI-compatible provider, exactly like the non-special-cased providers. No new backend class is required since Requesty is OpenAI-compatible passthrough.

Intentionally not touched (would be heavier / provider-specific and out of scope for a generic OpenAI-compatible LLM provider): the image-generation factory/model classes and the DB-seed ServiceProviderInitializer + related migration. Happy to add those too if desired.

Verification

• Typecheck: ran tsc --noEmit -p tsconfig.app.json in frontend/magic-web/packages/magic-admin. The added enum member, base-URL map entry, icon-map entry, and all mirrored map/config entries compile with zero new errors (verified by diffing the typecheck output against the unmodified baseline — the remaining errors pre-exist in unrelated files and are unaffected by this change).
• Live API check: 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 (gpt-4o-mini-2024-07-18), confirming the base URL, provider/model naming, and Bearer auth are correct.

Links

• https://requesty.ai
• https://docs.requesty.ai
• https://app.requesty.ai/api-keys

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.