feat: add Atomic Chat as first-class LLM provider

[yanalialiuk]

Summary

• Adds atomic-chat as a first-class provider in agent-core — OpenAI-compatible local API defaulting to http://127.0.0.1:1337/v1, keyless by default (atomic-chat-local placeholder token)
• Exposes Atomic Chat in the Next.js template: model picker, settings panel (optional ATOMIC_CHAT_BASE_URL / ATOMIC_CHAT_API_KEY), and env hydration
• Updates OpenAPI enum, README provider table, and doctor scripts for express/library templates

Usage

# agent-core / library / express
MODEL_PROVIDER=atomic-chat MODEL_ID=local FIRECRAWL_API_KEY=fc-... node ...

# optional overrides
ATOMIC_CHAT_BASE_URL=http://host.docker.internal:1337/v1
ATOMIC_CHAT_API_KEY=your-key

Model IDs come from GET /v1/models on the running Atomic Chat instance — pick any listed id in MODEL_ID or the Next UI.

Test plan

• pnpm test in agent-core — resolve-model tests pass (including new atomic-chat cases)
• Start Atomic Chat locally, select Atomic Chat in Next template settings, run an agent prompt
• Verify Docker setup with ATOMIC_CHAT_BASE_URL=http://host.docker.internal:1337/v1

Made with Cursor