Open Source · Self-Hosted · Production-Grade

The AI that remembers,
automates, and scales.

OpenSentinel is the open-source AI agent platform with 9 LLM providers, persistent memory, workflow automation, 5 specialist sub-agents, and SOC 2-ready enterprise security. Deploy on your infrastructure in minutes.

Install npm install opensentinel
opensentinel
$ |
300+ Features
124 Built-in Tools
9 LLM Providers
15+ Integrations
20 Agent Templates
6,400+ Unit Tests

AI assistants today are half-built.

They chat. They forget. They can't automate. They can't scale. They can't pass a security review. You deserve more than a chatbot with a GitHub repo.

Typical AI Assistants

  • Memory lost after every session
  • Manual chat only — no automation
  • Single agent, sequential tasks
  • No security controls or audit trail
  • Single user, personal device only
  • Prompt templates as "plugins"
  • No observability — flying blind

OpenSentinel

  • Agentic RAG: tool pre-classification, memory middleware, pipeline orchestrator
  • Advanced RAG: HyDE, re-ranking, multi-step retrieval, caching
  • Brain Dashboard: real-time pipeline visualization, activity feed, score gauges
  • Workflow engine with triggers & schedules
  • 5 specialist sub-agents in parallel
  • Gateway token auth, field encryption, tamper-proof audit
  • 2FA, incident response, GDPR, encrypted vault
  • Multi-user with SSO and quotas
  • Sandboxed plugins + MCP servers
  • Metrics, replay, alerting, dry-run, brain telemetry
  • Multi-LLM: 9 providers — Claude, OpenAI, xAI Grok, Gemini, Groq, Mistral, OpenRouter, Ollama, Custom
  • File download UI with secure token-based access
  • Image vision & document parsing from Web Chat uploads
  • ML algorithms: clustering, anomaly detection, forecasting
  • Finnhub & FRED financial data integrations
  • Prometheus metrics for DevOps monitoring
  • Built-in tunnels for instant public access

Everything you need to deploy production AI agents.

Not a framework. Not a wrapper. A complete platform with memory, tools, automation, and security built in.

Agentic RAG Memory

Full agentic RAG pipeline: tool pre-classification, memory middleware (auto-search + auto-extract), pipeline orchestrator, HyDE, re-ranking, multi-step retrieval, and Redis caching. Hybrid search fuses vector, keyword, and graph results. Your AI genuinely learns over time.

Core Differentiator

Workflow Automation

IFTTT-like engine with time, webhook, and event triggers. BullMQ scheduler for reliable execution. Build once, run forever.

Sub-Agent System

Five specialized agents — Research, Coding, Writing, Analysis, OSINT — with task coordination, shared context, and parallel execution.

Enterprise Security

2FA, biometric verification, encrypted vault, command sandboxing, GDPR compliance, comprehensive audit logging, and rate limiting.

15 Domain Experts

Coding, legal, medical, finance, marketing, design, data, security, DevOps, and more. Mood detection and tone adaptation built in.

File Generation & Parsing

Generate PDFs, Word docs, Excel spreadsheets, PowerPoint presentations, charts, diagrams, and AI images on demand. Secure download links with token-based access. Upload documents for automatic parsing (PDF, DOCX, CSV, and more). Image vision analysis from Web Chat uploads.

Multi-Channel Input

Telegram, Discord, Slack, WhatsApp, Signal, iMessage, WebSocket, web dashboard, REST API, voice wake word, and device triggers.

Full Observability

Brain Dashboard with real-time pipeline visualization, activity feed, and score gauges. ML-powered anomaly detection and cost forecasting. Plus metrics, conversation replay, tool dry-run, prompt inspector, cost alerting, and brain telemetry.

Plugin & MCP Support

Sandboxed plugin system with lifecycle management plus Model Context Protocol server support for connecting to external tool ecosystems.

Multi-Provider LLM

9 providers: Anthropic Claude, OpenAI, xAI Grok, Google Gemini, Groq, Mistral, OpenRouter, Ollama, or any OpenAI-compatible endpoint. Switch providers per-request or set a default. Run models locally for full privacy.

Autonomy Controls

Three autonomy levels — readonly, supervised, autonomous — control what the AI can do. Per-user settings, API control, and full audit trail.

ML Algorithms

5 built-in ML algorithms with zero external dependencies: Naive Bayes for intent classification, Isolation Forest for anomaly detection, K-Means clustering, Markov Chain prediction, and Linear Regression for cost forecasting and trend detection.

Prometheus Metrics

Standard Prometheus text format at /metrics. Request counters, token usage, latency histograms, and system gauges. Plug into Grafana instantly.

App Ecosystem SDK

Connect your applications to OpenSentinel's AI platform. TypeScript and Python SDKs with auto-registration, cross-app memory sharing, unified notifications, and sub-agent spawning.

  • TypeScript + Python SDKs
  • Growing app ecosystem
  • Cross-app intelligence
  • Unified notifications

Use as a library.
Build anything.

OpenSentinel isn't just a running application — it's an NPM package you can import into your own projects. Persistent memory, 124 tools, and enterprise security in one npm install.

  • Full TypeScript types included
  • Built on Bun, npm-installable
  • Tree-shakeable ESM exports
  • Lazy initialization — no env vars needed at import
app.ts
import { configure, chat, chatWithTools } from 'opensentinel';

// Configure with your API key
configure({
  CLAUDE_API_KEY: process.env.CLAUDE_API_KEY,
});

// Simple conversation
const response = await chat([
  { role: 'user', content: 'Summarize my inbox' }
]);

// With tools (browser, shell, files, search...)
const result = await chatWithTools([
  { role: 'user', content: 'Research competitors and create a report' }
]);

// Full platform with all services
import { OpenSentinel } from 'opensentinel';

const sentinel = new OpenSentinel({
  claudeApiKey: process.env.CLAUDE_API_KEY,
  telegram: true,
  discord: true,
  workflows: true,
});

await sentinel.start();

20 agent templates. Pick one. Ship today.

Each template is a standalone project that uses OpenSentinel as a library. Clone, configure your API key, and run.

🌐

Web Monitor

Monitor pages for changes, get intelligent alerts

💼

Sales Agent

Research leads, draft outreach, track pipeline

👥

Recruiter

Screen candidates, rank applicants, draft outreach

🖥

DevOps Agent

Server monitoring, log analysis, incident response

📈

Trading Researcher

Exchange trading, DeFi data, on-chain analytics, order books, backtesting

🎧

Customer Support

Ticket triage, response drafting, escalation routing

✍️

Content Creator

Multi-platform content from a single brief

🛡

Security Monitor

Auth log analysis, network audit, file integrity

🔍

Code Reviewer

PR review, bug detection, test coverage analysis

📊

Data Analyst

Dataset profiling, insights, anomaly detection

📧

Email Assistant

Inbox triage, reply drafting, action extraction

📝

Meeting Assistant

Transcript summaries, action items, weekly digests

🔎

Competitor Tracker

Product monitoring, pricing changes, hiring signals

🚀

SEO Optimizer

Page scoring, meta optimization, content outlines

⚖️

Legal Reviewer

Contract review, risk flagging, amendment suggestions

📢

Social Listener

Brand monitoring, sentiment analysis, trend detection

📖

Docs Writer

Auto-generate API refs, guides, changelogs

👋

Onboarding Agent

Personalized plans, Q&A, progress tracking

📦

Inventory Manager

Stock tracking, demand forecasting, purchase orders

🏠

Real Estate Analyst

Property analysis, market research, ROI estimation

Connects to everything you use.

Built-in integrations for communication, productivity, smart home, finance, cloud storage, and more.

🤖Claude
💬Telegram
🎮Discord
📨Slack
📱WhatsApp
🔒Signal
💭iMessage
📧Email Client
☎️Twilio
🐙GitHub
📓Notion
📅Google Cal
📆Outlook
🏠Home Assistant
🎵Spotify
📁Google Drive
📦Dropbox
💰Crypto
📈Stocks
🌍OpenAI
🎤ElevenLabs
🧠Ollama
🔌MCP Servers
🔗Webhooks
📣Matrix
🔄OpenRouter
Groq
🌪Mistral
💎Gemini
🚀xAI Grok
🤗HuggingFace
🔍OSINT
🏛Public Records
📊Finnhub
🏦FRED
💬Zalo
Cloudflare

Connected Applications

Build on the OpenSentinel platform with the App Ecosystem SDK. These applications leverage cross-app memory sharing, unified notifications, and AI-powered intelligence.

🎓TutorAI
📄DocGen AI
🛒EcomFlow
📊PolyMarketAI
🌿GoGreen Sourcing
TimeSheetAI
20+ more

Built for production, not demos.

PostgreSQL for persistent vector memory. Redis for reliable task queues. Docker and Kubernetes for deployment at any scale.

Inputs

Telegram Discord Slack WhatsApp Signal iMessage Matrix Web UI REST API Voice WebSocket

OpenSentinel Core

LLM Providers Agentic RAG Brain Telemetry Tool Classifier Sub-Agents (5) Workflow Engine Plugin System Scheduler Autonomy Prometheus

Outputs

Text Voice (TTS) PDF / Word Excel / PPT Charts AI Images Diagrams
PostgreSQL + pgvector Redis + BullMQ Docker / Kubernetes Tunnels (CF / ngrok)

The AI assistant your security team will approve.

Self-hosted on your infrastructure. No data leaves your network. SOC 2-ready controls, full audit trail, SSO integration, and compliance tools built in. Production-hardened Docker Compose with non-root containers, TLS, and resource limits.

2FA & Biometric Auth TOTP for sensitive operations, mobile biometric support
SSO Integration SAML, OAuth 2.0, and OpenID Connect
Tamper-Proof Audit Logging HMAC-SHA256 chain integrity, complete action history
GDPR Compliance Data export, deletion tools, retention policies
Usage Quotas & Rate Limiting Per-user controls, cost tracking, anomaly alerts
Kubernetes Native Helm charts, horizontal scaling, health checks
Security Dashboard
2FA Status Enabled
Memory Vault Encrypted
Audit Trail Recording
Command Sandbox Active
Plugin Isolation Enforced
GDPR Compliance Compliant
SOC 2 Readiness Ready
Threats Blocked (24h) 0

Running in under a minute.

Choose your path. Use OpenSentinel as a library in your project, or deploy the full platform.

Library

Use in Your Project

Add AI capabilities to any Bun application.

$ npm install opensentinel
import { configure, chat } from 'opensentinel';
configure({ CLAUDE_API_KEY: 'sk-ant-...' });
const res = await chat([...]);
Full Platform

Deploy Everything

One command installs everything: Bun, PostgreSQL, Redis, and OpenSentinel.

$ curl -fsSL https://opensentinel.ai/install.sh | bash
# Or install manually:
$ bun install -g opensentinel
$ opensentinel setup     # interactive wizard
Template

Start From a Template

Pick a use case, clone the template, and customize.

$ git clone https://github.com/dsiemon2/OpenSentinel.git
$ cd OpenSentinel/templates/ai-sales-agent
$ bun install
$ CLAUDE_API_KEY=sk-ant-... bun run start

Ready to deploy AI agents that actually work?

Open source. Self-hosted. Production-grade. Start building today.