Adoption Paths

Lexigram is designed for incremental adoption. You don’t need to install everything at once — start with the core and add packages as your application grows.

Tier 1 — Starter (1-3 packages)

The minimal foundation: dependency injection, application structure, and a Result type.

• lexigram + lexigram-contracts — Core: DI container, Application abstraction, Configuration, structured logging, Result type
• Add lexigram-web → Build an HTTP API with routing, middleware, request validation, and error handling
• Configuration via YAML files, structured logging built in

Tip

This tier is enough to build a complete API server. At this stage you have no database, no caching, and no auth — perfect for prototypes and early-stage services.

Tier 2 — Standard (4-8 packages)

Add persistence, caching, authentication, and developer tooling.

• Add lexigram-sql → Database persistence with migrations, repositories, and query building
• Add lexigram-cache → Response caching, session store, rate limiter backing
• Add lexigram-auth → JWT authentication, RBAC, route guards, password hashing
• Add lexigram-cli → Development server, database migrations, code scaffolding
• Add lexigram-testing → Fakes, test beds, and compliance test suites

Note

This tier is the sweet spot for most production applications. You have persistence, caching, auth, and testing — everything needed for a standard web application or API service.

Tier 3 — Advanced (8-15 packages)

Search, file storage, background jobs, event-driven architecture, notifications, and multi-tenancy.

• Add lexigram-search → Full-text search with Meilisearch, Elasticsearch, or Typesense
• Add lexigram-storage → File uploads, CDN integration, presigned URLs
• Add lexigram-tasks → Background job processing, cron scheduling, task queues
• Add lexigram-events → Event-driven architecture, CQRS, domain events
• Add lexigram-notification → Email, SMS, and push notification delivery
• Add lexigram-webhook → Outbound webhook dispatch with retry and signing
• Add lexigram-monitor + lexigram-resilience → Health checks, metrics, circuit breakers, retry policies
• Add lexigram-tenancy → Multi-tenant SaaS support with data isolation

Tip

The event bus (lexigram-events) and task queue (lexigram-tasks) both integrate with lexigram-resilience for fault tolerance. See Compatibility & Dependencies for these cross-package wiring notes.

Tier 4 — AI-Powered (15+ packages)

Add LLM integration, RAG, agents, memory, and production AI infrastructure.

• Add lexigram-ai-llm → LLM integration with provider routing (OpenAI, Anthropic, Google, local)
• Add lexigram-ai-rag + lexigram-vector → RAG pipeline with document ingestion, chunking, embedding, and hybrid search
• Add lexigram-ai-agents + lexigram-ai-skills → AI agents with tool use, multi-step reasoning, and sub-agent delegation
• Add lexigram-ai-memory + lexigram-ai-session → Persistent conversation history, episodic and semantic memory
• Add lexigram-ai-mcp → MCP server and client — expose agents as MCP tools, connect to external MCP servers
• Add lexigram-ai-workers + lexigram-ai-observability + lexigram-ai-feedback → Background processing, token usage tracking, cost attribution, feedback collection

Tip

You don’t have to commit to a tier all at once. Each package is independently installable and adopts incrementally. Mix and match — use lexigram-ai-llm without RAG, or lexigram-vector without the AI packages.

Decision Flow

Starting a new project?
├── Prototype or microservice → **Tier 1**
├── Standard web app or API → **Tier 2**
├── Scaling app with advanced needs → **Tier 3**
└── Building AI features → **Tier 2 first**, then **Tier 4**

Each tier is a superset of the one before it. You can move between tiers as your requirements evolve — no rewrite needed.

---

See Choosing Backends to pick the right infrastructure for each package, and Compatibility & Dependencies for version requirements and known constraints.