Skip to content

monorepo.md

Latest commit

 

History

History
57 lines (45 loc) · 2.44 KB

monorepo.md

File metadata and controls

57 lines (45 loc) · 2.44 KB
Raw
type infrastructure

Monorepo

The project is a pnpm workspaces monorepo with Turborepo for task orchestration. Seven deployables and shared libraries live in the same repository.

Layout

apps/                           # deployable processes
├── web/                        #   Next.js app (the UI)
├── worker/                     #   background worker
├── db-server/                  #   dev-only Postgres host
└── migrate/                    #   migration CLI + seed script

packages/                       # shared libraries
├── db/                         #   Prisma schema + client + apply-migrations
├── services/                   #   Azure adapters (queue, storage, notify)
├── domain/                     #   Zod schemas + queries (web-only)
├── log/                        #   pino logger
└── auth/                       #   dev identity stub

Dependency rules

  • Apps are deployable. Packages are shared.
  • No app imports from another app. The dependency graph flows through packages only. If two apps need the same utility, it goes in a package.
  • The @project scope is a placeholder. A real project replaces it via search-and-replace across all package.json files. The @project scope is never hardcoded in new code.

Key decisions

  • ADR-0001 — why monorepo, pnpm, Turborepo.
  • ADR-0009 — why packages/domain/ is web-only and the worker cannot import it.

Changing the layout

  • New package: add workspace to pnpm-workspace.yaml; create packages/<name>/; add "@project/<name>": "workspace:*" in the consuming app's dependencies.
  • New app: add workspace to pnpm-workspace.yaml; create apps/<name>/; add dependencies as needed.
  • Rename a package: update the dependency in every consumer's package.json; update all imports; update this doc and the relevant READMEs.

Skill payload (cache optimization)

Every opencode skill in .opencode/skill/ auto-loads its description into every session's system prompt and loads its body on demand (persisting for all subsequent turns). Body size directly increases conversation context per turn once loaded. When adding or editing a skill, prefer tight prose over template blocks — the actual files in docs/adr/ and docs/specs/ serve as canonical examples. This keeps prompt-cache costs stable as the skill set grows.