Skip to content

Providers

Jarett Schadlich edited this page Aug 28, 2026 · 1 revision

Providers

Quorum is model-agnostic. Claude, OpenAI, Meta/Llama, Kimi, and local / open-source models all plug in behind one adapter. OpenAI, Meta/Llama, Kimi, and local servers share one OpenAI-compatible HTTP path, so any OpenAI-compatible endpoint works too.

Pick the vendor and model when you seat an AI:

quorum agent lobby --provider anthropic --model claude-sonnet-5
# or, from inside a room:
/agent claude --provider openai --model gpt-5

List what's installable (with each one's models and required keys):

quorum providers

Add a model provider

Quorum is extensible at one seam: the provider adapter. To add a vendor, implement ProviderAdapter in a new file under src/providers/ and register it in src/providers/index.ts. The router, the room protocol, and the TUI never change — that's the whole point of the interface.

Repo layout

src/
  core/        CRDT live surface + DAG ledger
  relay/       self-hostable room server / chat backbone
  tui/         Ink terminal chat window
  router/      multi-model routing + delegation
  providers/   the model adapters — the extensibility surface
  cli.ts       command entry point

Develop

npm install
npm run dev -- providers     # run the CLI from source
npm run build                # compile to dist/
npm run typecheck
npm run test:all             # the full suite

See Publishing to npm for the release process.

Quorum (terminal + SDK)

Getting started

Run & share

Build

Design / RFCs

Maintainers


📱 The chat app is a separate repo: quorum-app

Clone this wiki locally