⚡ Open CLI — One CLI. Every AI. Zero Limits.

The AI terminal that thinks with you, builds with you, and ships with you.

opencli.myowncloud.tech — Built by Rythmm Costa

┌─────────────────────────────────────────────────────────────────┐
│  ⚡ Open CLI v2.0.0                              opencli.myowncloud.tech  │
├─────────────────────────────────────────────────────────────────┤
│                                                                   │
│  $ opencli "analyze project"                                      │
│  ⠋ Scanning 47 files...                                           │
│    TypeScript 5.x · Node.js · 18 modules                         │
│  ✓ Analysis complete (2.3s)                                       │
│                                                                   │
│  $ opencli --skill python "add REST API endpoint"                 │
│  ✓ Written to src/routes/users.py                                 │
│                                                                   │
│  $ opencli --ensemble "review this PR"                            │
│  ⠋ Querying Claude, GPT-4o, Gemini simultaneously...             │
│  ✓ 3 AI perspectives ready                                        │
│                                                                   │
│  $ opencli --free "explain this algorithm"                        │
│  ⠋ Using Groq llama-3.3-70b (free tier)...                       │
│  ✓ Explanation ready                                              │
│                                                                   │
└─────────────────────────────────────────────────────────────────┘

✨ What is Open CLI?

Open CLI is a professional AI terminal that unifies 17 AI providers — Claude, GPT-4o, Gemini, Groq, DeepSeek, Mistral, and more — behind a single, consistent command-line interface. Instead of juggling multiple web UIs, API clients, and workflows, you get one tool that integrates directly into your development environment.

At its core, Open CLI is a task execution engine: it reads your code, runs shell commands, manages files, automates browsers, queries databases, calls APIs, and interacts with GitHub — all under AI supervision. With 45+ domain-specific skills (Vue, React, Rust, Python, Docker, Kubernetes, ML, and more), it brings expert-level AI assistance precisely to the task at hand.

Open CLI also introduces Ensemble Mode, which queries multiple AI providers simultaneously and synthesizes their responses, giving you a broader and more reliable perspective on any problem. Combined with multi-window IPC, cross-device sync, and a free tier powered by Groq, Gemini Flash, and OpenRouter, Open CLI is the AI terminal designed for teams and developers who ship every day.

Installation

Option A — From GitHub (recommended)

git clone https://github.com/rythmmcosta/open-cli.git
cd open-cli
npm install && npm run build && npm link

Option B — NPM shortcut (Node.js 18+ required)

npm install -g github:rythmmcosta/open-cli

Windows

Use WSL2 for the best experience:

# In PowerShell as Administrator:
wsl --install
# After restart, in WSL2 terminal:
curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash -
sudo apt install -y nodejs git
git clone https://github.com/rythmmcosta/open-cli.git
cd open-cli && npm install && npm run build && npm link

Android (Termux)

pkg update && pkg install nodejs git
git clone https://github.com/rythmmcosta/open-cli.git
cd open-cli && npm install && npm run build && npm link

🚀 Quick Start

# Run interactive setup (adds API keys)
opencli auth

# Start coding with AI
opencli "explain the architecture of this codebase"

# Use a domain skill
opencli --skill react "create a data table component with sorting"

# Use only free AI tiers
opencli --free "write unit tests for this module"

# Query all AIs at once
opencli --ensemble "what's the best approach for this architecture?"

First-time setup takes about 2 minutes. You only need one API key to get started — or use the free tier with no key at all.

🧠 AI Providers (17 providers)

Configure providers interactively with opencli auth or by editing ~/.config/opencli/config.json.

Provider	Key Models	Free Tier	Setup
Anthropic (Claude)	claude-opus-4-5, claude-sonnet-4-5, claude-haiku-3-5	No	API key
OpenAI (GPT)	gpt-4o, gpt-4o-mini, o1-preview, o3-mini	No	API key
Google (Gemini)	gemini-2.0-flash, gemini-1.5-pro, gemini-1.5-flash	Yes (Flash)	API key
Groq	llama-3.3-70b, mixtral-8x7b, gemma2-9b	Yes (generous)	API key
DeepSeek	deepseek-chat, deepseek-reasoner	Yes (limited)	API key
Mistral AI	mistral-large-2, mistral-small, codestral	No	API key
Cohere	command-r-plus, command-r, command-light	Yes (trial)	API key
Perplexity	llama-3.1-sonar-large, sonar-pro	No	API key
xAI (Grok)	grok-2, grok-2-mini	No	API key
Kimi / Moonshot	moonshot-v1-8k, moonshot-v1-32k	No	API key
Together AI	llama-3.2-90b, qwen-2.5-72b	Yes (credits)	API key
Cerebras	llama3.1-70b, llama3.1-8b	Yes (fast free)	API key
HuggingFace	meta-llama/Llama-3.3-70B-Instruct, Qwen/Qwen2.5	Yes	API token
Azure OpenAI	gpt-4o, gpt-4o-mini (your deployment)	No	Endpoint + key
AWS Bedrock	claude-3.5-sonnet, llama3, titan-text	No	AWS credentials
Ollama	llama3.2, codellama, mistral, phi3 (local)	Free (local)	Local server
OpenRouter	200+ models including free tier	Yes (free models)	API key

Tip: Use opencli --free to automatically select from the free-tier providers. See docs/FREE-TIER.md for a complete guide.

🛠 Skills (45+ built-in skills)

Skills are deep, domain-specific AI personas that know the latest best practices, libraries, and patterns for their technology. Activate with opencli --skill <id> or /skill <id> inside the REPL.

Web Frontend

Skill ID	Name	Description
`javascript`	JavaScript	ES2024+, DOM, Web APIs, Node.js
`typescript`	TypeScript	Advanced types, generics, decorators, strict mode
`react`	React	React 18+, hooks, Next.js 14, Server Components
`nextjs`	Next.js 15	App Router, Server Actions, React 19, Turbopack
`vuejs`	Vue.js	Vue 3 Composition API, Pinia, Vue Router, Vite
`svelte`	Svelte 5 + SvelteKit	Runes reactivity, form actions, SSR
`html-css`	HTML & CSS	Semantic HTML5, modern CSS, container queries, animations
`tailwind`	Tailwind CSS	Tailwind 4.x, shadcn/ui, dark mode, design systems
`gsap`	GSAP	GSAP 3, ScrollTrigger, timelines, SVG animation
`threejs`	Three.js	WebGL, 3D scenes, shaders, react-three-fiber
`wasm`	WebAssembly	Wasm modules, Rust/C++ to Wasm, WASI

Backend

Skill ID	Name	Description
`nodejs`	Node.js	Express, Fastify, REST & GraphQL APIs, TypeScript
`python`	Python	Python 3.12+, FastAPI, async, data science
`php`	PHP	PHP 8.2+, Laravel, WordPress, Composer
`golang`	Go	Go 1.22+, goroutines, Gin/Echo/Fiber, table-driven tests
`rust`	Rust	Ownership, lifetimes, async, Tokio, axum
`java`	Java 21	Spring Boot 3.x, virtual threads, records, JUnit 5
`csharp`	C# / .NET	.NET 8+, ASP.NET Core Minimal APIs, EF Core 8
`kotlin`	Kotlin	Kotlin 2.x, coroutines, Ktor, Spring Boot
`swift`	Swift	Swift 5.9+, SwiftUI, async/await, Combine
`elixir`	Elixir	Phoenix, LiveView, OTP, pattern matching
`cpp`	C++	C++23, RAII, templates, performance optimization

Mobile

Skill ID	Name	Description
`mobile`	React Native	React Native 0.74+, Expo SDK 51, Expo Router v3
`flutter`	Flutter	Flutter 3.x, Dart 3, Riverpod 2, GoRouter, Material 3

Data & ML

Skill ID	Name	Description
`database`	Database	PostgreSQL, MySQL, MongoDB, Redis, SQL optimization
`sql`	SQL	Advanced SQL, window functions, query optimization
`ml`	Machine Learning	PyTorch 2.x, scikit-learn, HuggingFace Transformers
`r-lang`	R	Data analysis, ggplot2, tidyverse, statistical modeling
`graphql`	GraphQL	Schema design, Apollo Server 4, DataLoader, subscriptions
`prisma`	Prisma ORM	Schema, relations, migrations, Client queries

DevOps & Infrastructure

Skill ID	Name	Description
`docker`	Docker & DevOps	Multi-stage builds, Compose, CI/CD pipelines
`kubernetes`	Kubernetes	K8s 1.29+, Helm 3, ArgoCD GitOps, RBAC
`terraform`	Terraform	HCL, providers, modules, state, workspaces
`aws`	AWS	CDK v2 (TypeScript), Lambda, DynamoDB, S3
`cicd`	CI/CD	GitHub Actions, GitLab CI, ArgoCD, release automation
`nginx`	Nginx	Reverse proxy, SSL/TLS, rate limiting, caching
`bash`	Shell Scripting	Bash 5, POSIX sh, awk, sed, jq, parallel
`git`	Git	Workflows, branching strategies, GitHub, CI/CD

Security & Systems

Skill ID	Name	Description
`security`	Security	OWASP Top 10, secure coding, cryptography

Automation & AI

Skill ID	Name	Description
`browser-automation`	Browser Automation	Playwright: navigate, screenshot, verify, extract
`vibe-coding`	Vibe Coding	Creative AI-first development, rapid iteration

Web3 & CMS

Skill ID	Name	Description
`solidity`	Solidity / Web3	EVM smart contracts, Hardhat, ethers.js, OpenZeppelin
`wordpress`	WordPress	Gutenberg blocks, WP REST API, ACF, WooCommerce

Other

Skill ID	Name	Description
`regex`	Regex	Regular expressions in JS, Python, Go, PCRE
`testing`	Testing	Playwright E2E, Vitest, Jest, Testing Library

🔌 MCP Servers (26 built-in servers)

Open CLI ships with 26 built-in MCP (Model Context Protocol) servers. No external downloads required for core servers. Configure with opencli auth or /mcp list in the REPL.

Developer Tools

Server	Description	Setup Required
`github`	Repos, issues, PRs, file contents via GitHub REST API v3	GitHub PAT
`jira`	List/create/update issues, comments, transitions	Jira Cloud API token
`linear`	Issues, projects, cycles, team management	Linear API key
`slack`	Send messages, list channels, read channel history	Slack Bot token
`email`	Send emails via any SMTP server	SMTP URL
`clipboard`	Read/write system clipboard (macOS, Linux, Windows)	None

Data & Databases

Server	Description	Setup Required
`database`	SQLite: SELECT, INSERT, CREATE TABLE, migrations	None
`postgres`	PostgreSQL: query, schema inspection, migrations	Connection URL
`mongodb`	MongoDB: CRUD, aggregate, collections	MongoDB URI
`redis`	Redis: get/set/del, list, hash, pub-sub	Redis URL
`csv`	Parse, filter, aggregate, transform CSV files	None
`pdf`	Create PDFs, extract text from PDF files	None
`supabase`	Supabase database + auth + storage via REST API	Supabase URL + key

Cloud & Storage

Server	Description	Setup Required
`s3`	AWS S3: upload, download, list, delete objects	AWS credentials
`google-drive`	Google Drive: list, upload, download, share	OAuth credentials
`ftp`	FTP/SFTP: connect, list, upload, download	Host + credentials
`dropbox`	Dropbox: list, upload, download files	Dropbox API token

Integrations

Server	Description	Setup Required
`stripe`	Stripe: customers, payments, subscriptions, webhooks	Stripe secret key
`notion`	Notion: pages, databases, blocks, search	Notion integration token
`wordpress`	WordPress: posts, pages, media, users via REST API	WP URL + credentials
`image-gen`	Image generation: DALL-E, Stable Diffusion, Flux	Provider API key
`cli-tools`	Wrap any CLI tool as an MCP server	None

Core

Server	Description	Setup Required
`filesystem`	Read, write, list, and search files/directories	None
`shell`	Execute shell commands with safety scoring	None
`git`	Full git workflow: status, commit, push, diff, branches	None
`browser`	Playwright browser: navigate, click, screenshot, verify	`npx playwright install chromium`
`http`	HTTP/REST requests: GET, POST, PUT, DELETE	None
`search`	Web search via DuckDuckGo	None

💻 CLI Reference

Commands

opencli [prompt]                Chat mode with default model
opencli --skill <id>            Activate a domain skill
opencli --model <name>          Use a specific model
opencli --ensemble              Query all AIs simultaneously
opencli --free                  Use only free AI tiers
opencli --codex                 Coding-focused mode
opencli --image "prompt"        Generate an image
opencli --video "prompt"        Generate a video
opencli auth                    Manage AI providers (interactive)
opencli project                 Project management (init, list, switch)
opencli github                  GitHub operations (PR, issues, review)
opencli sync                    Cross-device sync (push, pull, status)
opencli workspace               Multi-window workspace status
opencli benchmark               Benchmark and compare models
opencli recipe                  Run/manage automation recipes
opencli timeline                File change history
opencli search                  Search conversation history
opencli costs                   Usage and cost dashboard

Flags

Flag	Short	Description
`--model <name>`	`-m`	Use specific AI model
`--skill <id>`	`-s`	Activate domain skill
`--ensemble`		Query all AI providers simultaneously
`--free`		Use only free AI tiers
`--codex`		Coding-focused mode with extra tools
`--image <prompt>`		Generate image from text
`--video <prompt>`		Generate video from text
`--dry-run`		Preview actions without executing
`--yes`	`-y`	Auto-approve low-risk commands
`--profile <name>`	`-p`	Use named config profile
`--show-cost`		Show token usage and cost
`--verbose`	`-v`	Show all tool calls
`--context <file>`	`-c`	Inject file into context
`--notify`		Send notification when done
`--browser`		Enable browser tools
`--no-mcp`		Disable all MCP servers
`--help`	`-h`	Show help
`--version`		Show version

REPL Commands

Inside the interactive REPL, use / commands:

/auth              Manage AI providers
/model <name>      Switch model mid-conversation
/skill <name>      Activate a skill
/skills            List all available skills
/ensemble          Toggle ensemble mode
/free              Toggle free-tier mode
/mcp list          List active MCP servers
/mcp enable <id>   Enable an MCP server
/mcp disable <id>  Disable an MCP server
/browser           Browser automation commands
/image <prompt>    Generate an image
/sync              Sync conversation across devices
/workspace         Show multi-window workspace
/history           Conversation history
/search <query>    Search past conversations
/timeline          File change timeline
/recipe <name>     Run a saved recipe
/cost              Token usage and cost report
/costs             Full cost dashboard
/benchmark         Run model benchmark
/github            GitHub operations
/project           Project context management
/reset             Clear conversation history
/config            Edit configuration
/help              Full command reference
/exit              Exit

🎯 AI Ensemble Mode

Ensemble Mode queries multiple AI providers simultaneously and presents their responses together — giving you diverse perspectives, consensus answers, and the ability to spot where models disagree.

# Query all configured AIs at once
opencli --ensemble "what's the best database for this use case?"

# Ensemble with only free providers
opencli --ensemble --free "review this code for bugs"

# Inside REPL
/ensemble on
> explain the trade-offs of microservices vs monolith

Why use Ensemble Mode?

Architecture decisions benefit from multiple expert perspectives
Code reviews catch issues that a single model might miss
Comparing models on your specific use case helps you choose the best one
Free ensemble mode costs nothing — all from free-tier providers

See docs/ENSEMBLE.md for the complete guide.

🪟 Multi-Window Workspace

Open CLI supports multiple terminal windows working on the same project simultaneously. Windows share context through a lightweight IPC (inter-process communication) layer with file locking to prevent conflicts.

# Check workspace status
opencli workspace

# In Window 1: long-running task
opencli "refactor the authentication module"

# In Window 2: parallel work continues safely
opencli --skill testing "write tests for the auth module"

# Windows automatically coordinate file access
# Conflicts are surfaced as warnings, not silent failures

Features:

File-level locking prevents concurrent write conflicts
Shared conversation context across windows (optional)
Workspace status shows which files each window is editing
Graceful handoff when a window exits

See docs/IPC-MULTIWINDOW.md for the complete guide.

☁️ Cross-Device Sync

Sync your Open CLI conversations, recipes, and settings across devices using your Open CLI account.

# Register or log in
opencli auth --account

# Push current session to cloud
opencli sync push

# Pull latest from cloud on another device
opencli sync pull

# Check sync status
opencli sync status

# Enable auto-sync
opencli sync enable --auto

What syncs:

Conversation history
Custom recipes
Skill configurations
Model preferences
Cost tracking data

Privacy: All sync data is encrypted at rest. Your API keys are never synced — they stay on each device.

See docs/SYNC.md for full details, quota limits, and account management.

🆓 Free AI Tier

Open CLI works 100% free using these providers:

Provider	Free Tier	Speed	Best For
Groq	Generous daily limit	Ultra-fast	General coding, Q&A
Gemini Flash	1M tokens/day	Fast	Long context, analysis
Cerebras	Daily limit	Ultra-fast	Quick completions
HuggingFace	Rate-limited	Variable	Specialized models
OpenRouter	Free model tier	Fast	Model variety
Ollama	Unlimited (local)	Local speed	Privacy, offline
DeepSeek	Limited daily	Fast	Reasoning tasks

# Use free providers automatically
opencli --free "explain this error"

# Ensemble across free providers only
opencli --free --ensemble "architecture review"

# Set free mode as default
opencli config set defaultFree true

See docs/FREE-TIER.md for setup guides and daily limits.

🖼 Image & Video Generation

Generate images and videos directly from the terminal:

# Generate an image
opencli --image "a futuristic terminal interface, dark theme, neon green"

# Generate a video (short clip)
opencli --video "code flowing through a neural network, cinematic"

# Inside REPL
/image dark fantasy landscape with glowing runes

Supported providers:

DALL-E 3 (OpenAI) — photorealistic and artistic
Stable Diffusion via Replicate — fine-grained control
Flux via fal.ai — high-quality, fast
Imagine via HuggingFace — free tier available

Images are saved to ~/.opencli/images/ and the path is printed to the terminal. In supported terminals (iTerm2, Kitty), images render inline.

⚙️ Configuration

Config file location: ~/.config/opencli/config.json

{
  "defaultModel": "claude-sonnet-4-5",
  "defaultSkill": "default",
  "defaultFree": false,
  "theme": "dark",
  "autoApprove": false,
  "autoApproveLevel": "low",
  "showCost": true,
  "showTokens": false,
  "maxTokens": 8192,
  "temperature": 0.7,
  "contextFiles": [],
  "profiles": {
    "work": { "defaultModel": "gpt-4o", "defaultSkill": "typescript" },
    "personal": { "defaultModel": "gemini-2.0-flash", "defaultFree": true }
  },
  "mcpServers": {
    "enabled": ["filesystem", "shell", "git", "browser", "http", "search"],
    "github": { "token": "" },
    "postgres": { "url": "" },
    "redis": { "url": "" },
    "slack": { "token": "" }
  },
  "notifications": {
    "telegram": { "enabled": false, "token": "", "chatId": "" },
    "discord": { "enabled": false, "webhookUrl": "" }
  },
  "sync": {
    "enabled": false,
    "autoSync": false
  }
}

Edit interactively:

opencli config          # Open config in $EDITOR
opencli config get      # Print current config
opencli config set defaultModel gpt-4o
opencli config set theme light

📖 Documentation

Document	Description
docs/INSTALLATION.md	Full setup guide, Node.js requirements
docs/PROVIDERS.md	All 17 AI providers: API keys, free tiers, models
docs/FREE-TIER.md	Use Open CLI 100% free
docs/ENSEMBLE.md	AI Ensemble Mode guide
docs/IPC-MULTIWINDOW.md	Multi-window workspace guide
docs/SYNC.md	Cross-device sync setup
docs/SKILLS.md	All built-in skills reference
docs/MCP-SERVERS.md	MCP servers reference
docs/BROWSER-AUTOMATION.md	Playwright browser integration
docs/AGENT-SYSTEM.md	Background agents setup
docs/NOTIFICATIONS.md	Telegram & Discord setup
docs/CONFIGURATION.md	Full config file reference
docs/CLI-FLAGS.md	All flags and commands
docs/SECURITY.md	Safety sandbox and security details

🤝 Contributing

Contributions are welcome! Open CLI is built with TypeScript 5.x and Node.js 20+.

# Clone and set up
git clone https://github.com/rythmmcosta/open-cli.git
cd open-cli
npm install
npm run build

# Link for local testing
npm link
opencli "test my local build"

# Run in development mode (watch)
npm run dev

How to contribute:

New Skill — Add a new Skill object in src/skills/definitions.ts. Each skill needs an id, name, description, systemPrompt, and examples.
New MCP Server — Add an MCPServer in src/mcp/registry.ts and implement tools in src/mcp/<name>.ts.
New AI Provider — Add provider config in src/providers/ and register in the provider registry.
Bug Fix / Enhancement — Open an issue first for significant changes. For small fixes, a PR is fine directly.

Code style: TypeScript strict mode, ESLint + Prettier. Run npm run lint before submitting.

See CONTRIBUTING.md for the full contributing guide.

📄 License

MIT — Rythmm Costa — opencli.myowncloud.tech

⚡ Open CLI — One CLI. Every AI. Zero Limits.

Website · npm · Issues · Discussions

Name		Name	Last commit message	Last commit date
.github		.github
docs		docs
src		src
website		website
.gitignore		.gitignore
README.md		README.md
package-lock.json		package-lock.json
package.json		package.json
tsconfig.json		tsconfig.json

Latest commit

History

Folders and files

Repository files navigation