feat(generic-openai): saved endpoint connections dropdown

[Souravrajvi0]

Pull Request Type

• ✨ feat (New feature)
• 🐛 fix (Bug fix)
• ♻️ refactor (Code refactoring without changing behavior)
• 💄 style (UI style changes)
• 🔨 chore (Build, CI, maintenance)
• 📝 docs (Documentation updates)

Relevant Issues

resolves #3871

Description

Adds saved Generic OpenAI endpoint profiles so users can store multiple base URLs / API keys / models (e.g. RunPod workers, LAN servers, LiteLLM routes) and switch between them from a dropdown in LLM Preference — without retyping settings each time.

Backend

• Persists profiles in storage/config/generic-openai-connections.json (gitignored; example file included)
• Admin API: list, save, activate, delete
• Selecting or saving a profile applies settings via existing updateENV path
• API keys are stored server-side only; list/summary endpoints never return secrets

Frontend

• Saved connection dropdown + Save / Delete actions on Generic OpenAI settings
• Applying a saved profile reloads the form with the active ENV values

This is a scoped first step toward the broader “stored connections” direction noted on #3871 / #2002 / #1567 (Generic OpenAI only for now).

Visuals (if applicable)

N/A — dropdown + buttons above existing Generic OpenAI fields in LLM Preference.

Additional Information

Local testing

yarn lint:ci
yarn test server/__tests__/utils/llm/genericOpenAiConnections.test.js
cd frontend && yarn build
cd docker && export UID=0 GID=0 && docker compose up -d --build

Results:

• yarn lint:ci — pass
• Connection manager unit tests — 4/4 pass (create/list summaries hide secrets, update preserves API key when omitted, activate via updateENV, delete)
• yarn build (frontend) — pass

Live API smoke test (yarn dev server + curl):

• List / save / activate / update / delete saved connections — pass
• setup-complete reflects applied Generic OpenAI fields after save/activate
• List API returns hasApiKey only (no raw apiKey field)
• Validation errors: missing name, unknown id, invalid URL — expected failures

Docker build & runtime (from docker/ per HOW_TO_USE_DOCKER.md):

• docker compose up -d --build — success (anythingllm-anything-llm:latest, ~4.67GB)
• Container anythingllm — healthy, http://localhost:3001/api/ping → {"online":true}
• Saved-connections API exercised inside running container — save + setup-complete pass

Developer Validations

• I ran yarn lint from the root of the repo & committed changes
• Relevant documentation has been updated (if applicable)
• I have tested my code functionality
• Docker build succeeds locally