โก From git clone to Full Company Operating System in 60 Seconds.
โฑ๏ธ Seriously. 60 Seconds.
You're a developer. You don't want to read a 40-page setup guide. You don't want to "contact sales." You don't want to configure 12 integrations before you see a dashboard.
You want to run a command and see something working.
That's exactly what HiveOps does.
git clone https://github.com/mamoor123/hiveops.git && cd hiveops
cp .env.example .env
openssl rand -base64 32 # paste into .env as JWT_SECRET
docker-compose up --build
Open http://localhost:3000. Done. ๐
You now have a full company operating system running on your machine. Let me show you what's inside. ๐ฆ
๐ง Wait โ What Did I Just Deploy?
A complete, production-ready platform that handles:
| Module | What It Does |
|---|---|
| ๐ Task Management | Create, assign, track โ with priorities, statuses, comments, and file attachments |
| ๐ค AI Agents | Per-department agents that autonomously execute tasks |
| โ๏ธ Workflow Engine | Triggers โ Conditions โ Actions. Your ops logic, automated |
| ๐ง Real Email | SMTP + IMAP. Inbox, drafts, replies. AI drafts responses for you |
| ๐ Knowledge Base | Articles with full-text search |
| ๐ฌ Real-Time Chat | Channel-based messaging with Socket.IO |
| ๐ Dashboard | Live stats on tasks, agents, emails, and workflows |
| ๐ Auth System | JWT tokens, role-based access, brute-force protection |
All in one Docker Compose file. ๐ณ
๐ What Happened in Those 60 Seconds?
Let me break down what docker-compose up actually started:
๐๏ธ 1. PostgreSQL 16
A full database with 14 tables โ ready to go.
users | departments | agents | tasks | task_comments | messages
knowledge_base | workflows | workflow_logs | emails
notifications | uploads | scheduled_tasks | schema_migrations
No migrations to run manually. The app handles it on boot. โ
๐ฅ๏ธ 2. The API Server (Node.js + Express)
Running on port 3001. Serving 12+ route modules:
# Auth
POST /api/auth/register โ Create account
POST /api/auth/login โ Get JWT token
GET /api/auth/me โ Current user
# Tasks
GET /api/tasks โ List tasks (filterable)
POST /api/tasks โ Create task
GET /api/tasks/:id โ Task detail
# AI Agents
POST /api/ai/chat/:agentId โ Chat with agent
POST /api/ai/execute/:taskId โ Execute task via agent
POST /api/ai/delegate โ Agent-to-agent delegation
# Workflows
POST /api/workflows โ Create workflow
POST /api/workflows/:id/toggle โ Enable/disable
# Email
GET /api/email/folder/:folder โ List emails
POST /api/email/send โ Send email
# Knowledge Base
POST /api/knowledge/search โ Full-text search
๐จ 3. The Frontend (Next.js 14)
Running on port 3000. App Router. 12 pages covering every module.
๐ค The AI Agents โ This Is Where It Gets Interesting
Most "AI integrations" are chat widgets. Type a question, get an answer. That's it.
HiveOps agents are workers. ๐ช
๐ข One agent per department
๐ฃ Marketing Agent
โ system prompt: "You handle campaigns, content, and social media strategy"
โ receives tasks assigned to marketing
โ executes them autonomously
๐ป Engineering Agent
โ system prompt: "You handle code reviews, deployments, and bug triage"
โ receives tasks assigned to engineering
โ executes them autonomously
๐ง Support Agent
โ system prompt: "You handle customer issues, escalations, and follow-ups"
โ receives tasks assigned to support
โ executes them autonomously
โก The execution loop
๐ฅ Task assigned to agent
โ ๐ค Agent picks it up
โ โถ๏ธ Executes with the task context
โ โ
Success? Mark complete.
โ โ Failed?
โ ๐ Retry #1 (10s delay)
โ โ Failed?
โ ๐ Retry #2 (20s delay)
โ โ Failed?
โ ๐ Retry #3 (40s delay)
โ โ Failed?
โ ๐ Dead letter queue
โ ๐ Notify task creator
Exponential backoff. Dead letter queues. Error history tracking. This isn't a weekend project. This is production infrastructure. ๐๏ธ
๐ Agent delegation
Agent A can hand off a sub-task to Agent B:
๐ค Engineering Agent: "I need a deployment verification"
โ delegates to โ
๐ค DevOps Agent: "Running smoke tests..."
โ reports back โ
๐ค Engineering Agent: "All clear. Marking task complete."
Chain agents together for complex workflows. ๐
โ๏ธ Workflow Engine โ Zapier Without the Bill
Build automation rules with three components:
๐ Triggers (when something happens)
-
task_createdโ A new task is born -
task_completedโ A task is done -
task_updatedโ Someone changed a field -
user_registeredโ New team member -
schedule_dailyโ Runs every day
๐ฏ Conditions (only if...)
priority == urgent-
due_dateis past -
field containsspecific text status == blocked- A field
exists
โถ๏ธ Actions (then do this)
- ๐ Notify a user, manager, or admin
- ๐ Update task fields
- ๐ฌ Send a channel message
- โ Create a new task
Example workflow:
Trigger: task_created
Condition: priority == urgent
Action: notify manager
Trigger: schedule_daily
Condition: due_date is past
Action: notify assigned_user + update status to "blocked"
Trigger: task_completed
Condition: department == engineering
Action: create new task "Deploy to staging" for DevOps agent
Build it once. It runs forever. No per-action billing. No premium tier. ๐
๐ง Email That Actually Works
Not a "send notification" endpoint. Real email. ๐ฌ
- ๐ฅ Inbox โ pull emails via IMAP polling
- ๐ค Send โ SMTP outbound (Gmail, custom SMTP)
- ๐ Drafts โ AI writes draft replies based on email context
- โญ Star & Label โ organize your mail
- ๐ Search โ find anything
- ๐ Notifications โ new email? You know instantly
The agent reads incoming emails and drafts intelligent responses. Review and send. Or let it send autonomously โ your call. โ๏ธ
๐ ๏ธ Under the Hood
Backend Architecture
server/
โโโ src/
โ โโโ config/
โ โ โโโ db.js ๐๏ธ Dual SQLite/PostgreSQL adapter
โ โ โโโ logger.js ๐ Pino structured logging
โ โ โโโ migrate.js ๐ Versioned migrations
โ โโโ middleware/
โ โ โโโ auth.js ๐ JWT + role-based access
โ โ โโโ rateLimit.js ๐ก๏ธ Brute-force protection (login abuse)
โ โ โโโ validate.js ๐งน Input sanitization
โ โโโ routes/ ๐ฃ๏ธ 12 API modules
โ โโโ services/
โ โโโ ai-engine.js ๐ง LLM integration (OpenAI-compatible)
โ โโโ email-real.js ๐ง SMTP + IMAP
โ โโโ execution-loop.js ๐ Retry + exponential backoff + DLQ
โ โโโ scheduler.js โฐ DB-persisted cron jobs
โ โโโ workflows.js โ๏ธ Rule engine
โ โโโ notifications.js ๐ Socket.IO broadcast
โโโ __tests__/ ๐งช 58 tests (Jest + Supertest)
Frontend
web/
โโโ app/ ๐ 12 pages (Next.js 14 App Router)
โโโ components/ ๐งฉ Shared UI components
โโโ lib/ ๐ API client + auth helpers
Tests โ Because We're Serious
| Suite | Tests | Covers |
|---|---|---|
| ๐ Auth | 14 | Register, login, profile, password, tokens |
| ๐ Tasks | 13 | CRUD, filters, comments, completion |
| โ๏ธ Workflows | 8 | CRUD, toggle, triggers, validation |
| ๐ค AI | 11 | Chat, execution, delegation, agents |
| ๐ข Departments | 12 | Departments, knowledge, email |
58 tests. All passing. โ
๐๏ธ Database: SQLite or PostgreSQL โ Your Choice
Same API. Same code. One env var difference.
# SQLite (default โ just works)
DB_PATH=./data/hiveops.db
# PostgreSQL (production scale)
DATABASE_URL=postgres://user:pass@host:5432/hiveops
| Mode | Best For |
|---|---|
| ๐ข SQLite | Local dev, prototyping, small teams |
| ๐ต PostgreSQL | Production, scale, complex queries |
Switch anytime. No code changes. No migrations to redo. ๐
๐ง LLM: Works With Anything OpenAI-Compatible
# OpenAI
LLM_API_URL=https://api.openai.com/v1/chat/completions
LLM_API_KEY=sk-your-key
# Ollama (local)
LLM_API_URL=http://localhost:11434/v1/chat/completions
LLM_API_KEY=ollama
# No key? No problem.
# Simulated fallback mode works out of the box.
No vendor lock-in. Run your own models locally with Ollama. Or plug in any provider. ๐
๐ The Numbers That Matter
| Metric | Value |
|---|---|
| ๐ณ Docker containers | 3 (Postgres + API + Web) |
| โฑ๏ธ Setup time | ~60 seconds |
| ๐ API routes | 12 modules |
| ๐๏ธ Database tables | 14 |
| ๐งช Test suite | 58 tests |
| ๐ฐ Cost | $0 forever |
| ๐ Data location | Your server |
๐ฏ Who Should Use This?
- ๐ Startups โ Can't burn $500/mo on SaaS before you have revenue
- ๐ Remote teams โ One platform instead of five disconnected tabs
- ๐ฉโ๐ป Developers โ Self-hosted > vendor-dependent
- ๐๏ธ Agencies โ White-label the OS for your clients
- ๐ค Anyone with SaaS fatigue โ Time to own your stack
๐ฎ What's Next?
The project is alive and growing. Here's what's on the roadmap (based on the codebase):
- ๐ More dashboard widgets
- ๐ Additional LLM provider integrations
- ๐ฑ Mobile-responsive improvements
- ๐ท๏ธ Enhanced labeling and tagging
- ๐ Reporting and analytics
๐ Open Source. Fully. No Gimmicks.
No "open core." No "enterprise edition." No "contact sales." ๐ซ
git clone https://github.com/mamoor123/hiveops.git
โญ Star it if you like it. ๐ด Fork it if you want to change it. ๐ Deploy it if you need it.
๐ github.com/mamoor123/hiveops
๐ Quick Links
| Link | Description |
|---|---|
| ๐ README | Full documentation |
| ๐ Issues | Bugs & feature requests |
| ๐งช Tests | 58 tests |
| ๐ณ Docker | Compose file |
๐ From git clone to full company OS. One command. 60 seconds. Zero dollars.
Tags: opensource, selfhosted, ai, docker, devops, productivity, nodejs, nextjs
Related Reading
- I Was Paying $600/mo for 5 SaaS Tools. So I Built One That Does All of Them.
- I Open-Sourced a Company OS That Replaces 5 SaaS Tools
- Your Data. Your Server. Your Agents. Zero SaaS Bills.
- I Sent One Message and 5 AI Agents Built, Audited, Tested & Deployed a Full App
- 10 Docker Commands That Actually Matter in 2026
Top comments (1)
Thanks