Beginner Issue: Add a remote-backend quickstart for Ollama, LM Studio, and OpenAI-compatible endpoints

[ambicuity]

Type

Documentation

Area

Remote backend and model integration

Difficulty

Beginner

Maintainer Guidance Needed

Medium

Problem

The repository supports a remote LLM backend and even has tests covering remote embedding/query flow, but the README only exposes environment variables. There is no quickstart showing how to point KINDX at Ollama, LM Studio, or another OpenAI-compatible endpoint.

Why It Matters

Without a quickstart, users cannot easily discover or trust a major fallback path for machines where local model execution is not desirable or not possible.

Scope

Add a short remote-backend guide to the README that covers backend selection, the required environment variables, optional API keys, and how rerank fallback behaves when /v1/rerank is unavailable.

Acceptance Criteria

• The README shows a minimal remote-backend setup example.
• The docs mention KINDX_LLM_BACKEND=remote and the required endpoint variables.
• The docs explain the current rerank fallback behavior for backends without /v1/rerank.
• The examples are consistent with the existing remote-backend tests.

Relevant Files

• README.md
• engine/remote-llm.ts
• specs/command-line.test.ts

Testing

• Cross-check the docs against engine/remote-llm.ts defaults and behavior.
• Run npm run build.

Non-goals

• Adding new remote-backend features
• Implementing provider-specific adapters
• Writing a full hosted deployment guide

Difficulty Rationale

This is still documentation work, but it requires reading the remote backend implementation and tests closely enough to avoid inaccurate examples.

Checklist

• I searched open and recently closed issues before drafting this task.
• The issue is scoped to one focused PR.
• The likely files and verification steps are listed.

---

Intermediate Issues