npm install @deepagents/agent
npm install @deepagents/text2sql
npm install @deepagents/context

npm install @deepagents/agentnpm install @deepagents/text2sqlnpm install @deepagents/context

import { agent, execute } from '@deepagents/agent';
import { groq } from '@ai-sdk/groq';

const assistant = agent({
  name: 'Assistant',
  model: groq('gpt-oss-20b'),
  prompt: 'You are a helpful assistant.',
});

await execute(assistant, 'Hello!', {});

import { agent, execute } from '@deepagents/agent';import { groq } from '@ai-sdk/groq';const assistant = agent({  name: 'Assistant',  model: groq('gpt-oss-20b'),  prompt: 'You are a helpful assistant.',});await execute(assistant, 'Hello!', {});

import { Text2Sql, InMemoryHistory } from '@deepagents/text2sql';
import { Postgres } from '@deepagents/text2sql/postgres';

const text2sql = new Text2Sql({
  version: 'v1',
  adapter: new Postgres({ execute, grounding: [] }),
  history: new InMemoryHistory(),
});

await text2sql.toSql('Show all customers');

import { Text2Sql, InMemoryHistory } from '@deepagents/text2sql';import { Postgres } from '@deepagents/text2sql/postgres';const text2sql = new Text2Sql({  version: 'v1',  adapter: new Postgres({ execute, grounding: [] }),  history: new InMemoryHistory(),});await text2sql.toSql('Show all customers');

import { ContextEngine, role, user, XmlRenderer } from '@deepagents/context';
import { SqliteContextStore } from '@deepagents/context';

const context = new ContextEngine({
  store: new SqliteContextStore('./chat.db'),
  chatId: 'session-1',
  userId: 'user-1',
});

context.set(role('You are helpful'), user('Hello'));
await context.resolve({ renderer: new XmlRenderer() });

import { ContextEngine, role, user, XmlRenderer } from '@deepagents/context';import { SqliteContextStore } from '@deepagents/context';const context = new ContextEngine({  store: new SqliteContextStore('./chat.db'),  chatId: 'session-1',  userId: 'user-1',});context.set(role('You are helpful'), user('Hello'));await context.resolve({ renderer: new XmlRenderer() });

import { ingest, fastembed, nodeSQLite } from '@deepagents/retrieval';
import * as connectors from '@deepagents/retrieval/connectors';

await ingest({
  connector: connectors.local('./docs'),
  store: nodeSQLite('./vectors.db', 1024),
  embedder: fastembed(),
});

import { ingest, fastembed, nodeSQLite } from '@deepagents/retrieval';import * as connectors from '@deepagents/retrieval/connectors';await ingest({  connector: connectors.local('./docs'),  store: nodeSQLite('./vectors.db', 1024),  embedder: fastembed(),});

DEEP
AGENTS

AI-native building blocks for production systems. Composable packages for agents, retrieval, and natural language interfaces.

Get Started View Source

PACKAGES

@deepagents/agent

Core agent framework with tool orchestration and multi-provider support.

-->

@deepagents/text2sql

Natural language to SQL with domain learning and teachable context.

-->

@deepagents/context

Context management with tokenization and persistence layer.

-->

@deepagents/retrieval

Vector & semantic retrieval with embeddings and full-text search.

-->

@deepagents/orchestrator

Multi-agent workflow coordination and task distribution.

-->

@deepagents/toolbox

Common utilities: web search, shell execution, and helpers.

-->

WHY DEEP AGENTS

const features = {
  production: "Built for real workloads",
  composable: "Use only what you need",
  batteryPowered: "Built-in retrieval, text2sql, and tools",
};

const features = {  production: "Built for real workloads",  composable: "Use only what you need",  batteryPowered: "Built-in retrieval, text2sql, and tools",};

Production-Ready

Built for real workloads with proper error handling, retries, and observability.

Composable

Mix and match packages. Use only what you need. No monolithic frameworks.

Battery Powered

Built-in retrieval, embeddings, connectors, and tools. Everything you need out of the box.

QUICK START

import { agent, execute } from '@deepagents/agent';
import { similaritySearch, nodeSQLite, fastembed } from '@deepagents/retrieval';
import * as connectors from '@deepagents/retrieval/connectors';
import { groq } from '@ai-sdk/groq';

const results = await similaritySearch(query, {
  store: nodeSQLite('./knowledge.sqlite', 1024),
  embedder: fastembed(),
  connector: connectors.github.repo(
    'https://github.com/JanuaryLabs/deepagents',
    { includes: ['docs/**/*.md'] }
  ),
});

const assistant = agent({
  name: 'Assistant',
  model: groq('gpt-oss-20b'),
  prompt: results.map(r => r.content).join('\n'),
});

await execute(assistant, "What can DeepAgents do?");

import { agent, execute } from '@deepagents/agent';import { similaritySearch, nodeSQLite, fastembed } from '@deepagents/retrieval';import * as connectors from '@deepagents/retrieval/connectors';import { groq } from '@ai-sdk/groq';const results = await similaritySearch(query, {  store: nodeSQLite('./knowledge.sqlite', 1024),  embedder: fastembed(),  connector: connectors.github.repo(    'https://github.com/JanuaryLabs/deepagents',    { includes: ['docs/**/*.md'] }  ),});const assistant = agent({  name: 'Assistant',  model: groq('gpt-oss-20b'),  prompt: results.map(r => r.content).join('\n'),});await execute(assistant, "What can DeepAgents do?");

DEEPAGENTS

PACKAGES

WHY DEEP AGENTS

QUICK START

DEEP
AGENTS