Developer documentation

Everything you need to connect, import data, and start querying with LoamDB.

Get running in 3 steps

From zero to your first query in under 5 minutes.

Connect to LoamDB

import { createLoamDB } from '@loamdb/client';

const db = createLoamDB({
  projectId: 'your-project-id',
  apiKey: process.env.LOAMDB_API_KEY,
});

Import your data

// Upload a CSV — LoamDB auto-detects schema
const job = await db.import({
  file: 'people.csv',
  type: 'auto', // or 'people', 'clients', 'deals'
});

// Check import status
const status = await db.importStatus(job.id);

Query your data

// Direct lookup
const person = await db.lookup('alice@company.com');

// Semantic search
const results = await db.search('engineering managers');

// Graph traversal
const reports = await db.traverse(person.id, {
  direction: 'down',
  depth: 3,
});

MCP tool reference

17 tools available via the Model Context Protocol for structured agent interactions.

Tool	Description
query_entities	Query any entity type with filters, sorting, and pagination
get_entity	Retrieve a single entity by ID with full metadata
search_semantic	Vector similarity search across all entity types
traverse_graph	BFS/DFS traversal of relationship graphs
get_org_chart	Retrieve org structure for a given root entity
import_csv	Ingest CSV data with automatic schema detection
import_status	Check status of an ongoing import job
manage_permissions	Create, update, and delete permission rules
check_access	Verify if a user can access a specific entity
get_schema	Retrieve the current database schema and entity types
list_imports	List all import jobs with status and metadata
get_statistics	Database-level stats: row counts, storage, index health
create_entity	Insert a new entity with validation and indexing
update_entity	Update entity fields with audit logging
delete_entity	Soft-delete an entity with cascade handling
bulk_operations	Batch create, update, or delete operations
export_data	Export query results in CSV, JSON, or Parquet format

Function-calling tool reference

7 tools optimized for LLM function-calling patterns with latency benchmarks.

Tool	Description	Latency
lookup_person	Direct entity lookup by email or name	~20ms
search_knowledge	Semantic search with relevance scoring	~150ms
get_org_chart	Manager hierarchy traversal	~40ms
query_pipeline	Deal pipeline queries with stage filters	~30ms
check_permissions	Verify access before data retrieval	~5ms
get_client	Client profile with deal history	~25ms
generate_report	Full RAG synthesis across multiple sources	~600ms

Data model overview

Four data models unified on Postgres. Each entity can be accessed through any model.

Relational

Drizzle ORM with TypeScript type safety. Normalized tables for people, clients, deals, compensation.

Graph

Adjacency-list storage with BFS traversal. Org charts, reporting lines, team hierarchies.

Vector

pgvector with HNSW indexing. Semantic search across all entities. 1536-dimension embeddings.

Document

JSONB columns with Schema.org alignment. Flexible metadata per entity type.

Full documentation coming soon

Comprehensive API docs, SDK references, and integration guides will be available at docs.loamdb.com.

Ready to get started?

Start Free Explore Features