ait
Knowledge pack for AIt — a personal AI platform with monorepo architecture, RAG pipeline, OAuth connectors, and BullMQ scheduling.
SKILL.mdskills/claudskills/ait/SKILL.md
GitHubait
Personal AI platform built as a pnpm monorepo. Connects to external services (GitHub, Spotify, Linear, Notion, Slack, Google) via OAuth, syncs data through an ETL pipeline, stores embeddings in Qdrant, and provides RAG-powered AI responses via Ollama.
Preferences
- Package manager: pnpm (corepack, workspace protocol)
- Linter/formatter: Biome 1.9 — 2-space indent, single quotes, semicolons
- Error handling:
Result<T, E>pattern withok()/err()— never throw in business logic - Testing: Node.js native
node:testwithdescribe/it,borprunner,c8coverage - Naming: PascalCase classes,
Iprefix interfaces, camelCase methods, SCREAMING_SNAKE constants, kebab-case files - Type files:
.types.ts, spec files:.spec.ts - All packages: scoped
@ait/*
References
| Category | Reference | Description |
|---|---|---|
| Core | core-architecture | Layered architecture, data flow, package structure |
| Core | core-entity-types | Normalized entity types with __type strings |
| Core | core-result-type | Result<T,E>, ok(), err() usage patterns |
| Features | features-oauth | Encrypted credentials, factory pattern, token refresh |
| Features | features-rag | Qdrant, multi-query retrieval, reranking, streaming |
| Features | features-scheduler | BullMQ, priority queues, repeatable jobs |
| Features | features-connectors | GitHub, Spotify, Linear, Notion, Slack, Google connectors |
| Best Practices | best-practices-testing | Node.js native test runner, Docker services, c8 coverage |
Quick Reference
# Development
pnpm dev # Start all packages in parallel
pnpm start:services # Docker: PostgreSQL, Qdrant, Redis, Ollama, MinIO
pnpm test # Run all tests (starts test services automatically)
pnpm migrate # Run Drizzle migrations
pnpm generate:openapi # Generate OpenAPI specs for connectors
# Ports
# PostgreSQL: 5432 | Qdrant: 6333 | Redis: 6379
# Ollama: 11434 | MinIO: 9090 | Langfuse: 3333
# Gateway: 3000 | UIt (UI): 5173
# Models (env)
GENERATION_MODEL=gemma3:latest
EMBEDDINGS_MODEL=mxbai-embed-large:latest
// Result pattern
const result = await someOperation();
if (result.ok) {
console.log(result.value);
} else {
console.error(result.error);
}
// Connector factory
const service = connectorServiceFactory.getServiceByConfig(configId, userId);