lounge.

Lounge Wiki

Lounge is a platform for building backend applications as trees: you describe an application's routing and task nodes in project.yaml, and Lounge executes events through it — sessions, queries, persistence, concurrency, and operations included. The description is executable as written; the same spec runs on a laptop, a server, or a cluster. (Naming in one line: you write an AppTree, Lounge runs it — a LoungeTech product. The engine's pre-rebrand name, Lounge, survives in lounge.* config keys until the code rename lands.)

Start by building a first application. Then explore the concepts below.

Pages

Page What it answers
The Tree and the Flow Start here. What Lounge is, the whole engine on one screen, then the walk an event takes: nodes, stepwise routing, exceptions, and the membrane
Event Groups, Ordering and Parallelism Which events wait for each other, which run side by side, and what "in order" promises
Records and Queries What the tree remembers and how you ask about it: records, JEQL, indexes, retention, the cold tier
Subtrees, Mounts and Scopes Reusable pieces, runtime composition, which store a query reads, giving a subtree its own front door, and running parts elsewhere
Integrations Talking to the outside world: databases, HTTP, mail, storage, models — and where the line sits
Durability, Clustering and HA Surviving a restart, running a subtree in its own process, running on more than one machine, handing over an identity
Building, Operating and Securing The work around the app: the linter and the portal, logs and surfaces, auth, protocols and hardening
Editions What CORE, PRO and CLUSTER include, how capabilities are enforced, and how billing works
Developing with LLMs Asking a model for a specification rather than code: records first, the refinement gate, and how far a small local model gets
Recipes Cut and paste. Build on metal or as an image, configure, Supabase vs plain Postgres, durability, archiving
Performance evidence Measured workloads, correctness checks and limits
CORE source and local model CORE source status, published model download, local setup and licenses
FAQ The questions everyone asks first

Deep documentation

The wiki is the map; the repo docs are the territory:

  • Getting started: docs/guide/getting-started.md
  • The spec language: docs/guide/project-yaml.md
  • Queries (JEQL): docs/guide/queries.md, docs/jeql/
  • Persistence: docs/guide/persistence.md
  • Architecture internals: docs/architecture/membrane-flows.md, docs/architecture/storage-and-query.md
  • Operations: docs/ops/ (auth, logging, clustering, wire, HA recovery)
  • Product: docs/product/tier-sheet.md