github.com/svenknobloc...
clawdbot
cursor.sh
cursor
codeium.com/windsurf
windsurf
docs.anthropic.com/en/...
claude-code
github.com/openai/codex
openai-codex-cli
aider.chat
aider
Clawdbot is a multi-channel AI assistant that runs across Telegram, Discord, WhatsApp, and more. Unlike chatbots trapped in web UIs, Clawdbot meets you where you already are. Built for builders who want an AI that: - Runs shell commands and manages files - Controls your browser - Integrates with your actual workflow - Works across all your devices It's Claude with hands - and it knows how to use them. Self-hosted, extensible, and unapologetically powerful.
0
open-source
Cursor is VSCode rebuilt from the ground up for AI-assisted coding. It's not a plugin - it's a completely reimagined editor where AI is a first-class citizen. The killer features: - Cmd+K to edit code with natural language - Codebase-aware chat that understands your entire project - AI-powered autocomplete that actually gets your style - Tab to accept inline suggestions If you're still using Copilot in regular VSCode, you're missing out. Cursor is what happens when AI isn't bolted on but built in.
freemium
Windsurf by Codeium takes AI coding beyond autocomplete. It's an IDE where AI doesn't just suggest - it flows with you. What sets it apart: - Cascade: An agentic mode that tackles multi-file changes - Deep codebase awareness across your entire project - Built on VSCode but engineered for AI-native workflows - Free tier that's actually usable While Cursor charges $20/month, Windsurf offers serious competition with a generous free plan. It's the underdog worth betting on.
Claude Code is Anthropic's answer to terminal-based AI coding. It's Claude, but wired directly into your shell. What it does well: - Full agentic coding in your terminal - Reads and writes files autonomously - Runs shell commands with approval - Understands your project context deeply No VS Code required. If you believe the future of coding is conversational, Claude Code is the purest expression of that vision. Available through Claude Max subscription.
subscription
Codex CLI brings OpenAI's coding models directly to your terminal. It's the official CLI tool for developers who prefer shells over GUIs. Core capabilities: - Execute coding tasks from natural language - File system awareness and manipulation - Shell command generation and execution - Works with GPT-4 and GPT-4o models OpenAI's bet on agentic coding. Still early, but improving rapidly. Requires API credits.
api-credits
Devin by Cognition Labs made headlines as the first "AI software engineer." It's not an assistant - it's designed to be a full teammate. What Devin can do: - Plan and execute entire coding tasks - Debug its own code through iteration - Learn unfamiliar technologies on the fly - Work autonomously for hours The demo videos are impressive. Real-world results vary. But this is the direction we're heading: AI that doesn't just help you code, but codes alongside you.
waitlist
GitHub Copilot is the OG AI coding assistant. Trained on billions of lines of code, it autocompletes your thoughts before you finish typing them. The feature set: - Inline code suggestions as you type - Chat interface in VS Code and JetBrains - Copilot Workspace for agentic multi-file edits - Deep GitHub integration (issues, PRs, code review) It's the most widely adopted AI coding tool. Not the most powerful anymore, but the most polished. $10/month for individuals.
AutoGPT kicked off the autonomous agent craze. Give it a goal, and it breaks it down into tasks, executes them, and iterates until done. The evolution: - Started as a viral GitHub experiment - Now a full platform for building agents - Agent Builder for no-code agent creation - Marketplace for sharing and monetizing agents The original vision of "AI that does things for you" keeps getting refined here. Open source at its core.
LangChain is the framework that powers most production LLM apps you've used. It's the plumbing behind the magic. What it provides: - Chains: Connect LLM calls with logic - Agents: LLMs that decide what actions to take - RAG: Retrieval-augmented generation made easy - Memory: Persistent context across conversations If you're building anything serious with LLMs, you'll probably touch LangChain. Python and JS/TS support. Huge ecosystem.
CrewAI lets you build teams of AI agents that work together. Think of it as project management for autonomous AI. The core concept: - Define Agents with specific roles and goals - Create Tasks they need to complete - Let them collaborate and hand off work - Watch the crew get things done It's LangChain but focused entirely on multi-agent systems. When one agent isn't enough, CrewAI is how you scale. Python-first, very active development.
Aider is the terminal-first AI coding tool that actually works. No IDE required - just your shell and your repo. Why developers love it: - Git-aware: commits changes automatically - Works with Claude, GPT-4, and local models - Edits multiple files coherently - Benchmarks consistently top the charts Free, open source, and absurdly capable. If you prefer vim/neovim or just love the terminal, Aider is your jam.
Continue is the open-source alternative to Copilot and Cursor. Full control, any model, no vendor lock-in. Key differentiators: - Use any LLM: Claude, GPT-4, Llama, Ollama, etc. - VS Code and JetBrains extensions - Custom slash commands and context providers - Self-host or use their cloud For teams that need flexibility or can't send code to third parties, Continue is the answer. Apache 2.0 licensed.
Replit Agent turns prompts into deployed apps. Describe your idea, watch it build, ship in minutes. The magic: - Creates full projects from descriptions - Sets up databases, auth, APIs automatically - Deploys to Replit hosting instantly - Iterates based on your feedback It's the closest thing to "vibe coding" we have. Great for prototypes and MVPs. Less control, more speed.
v0 generates React components from natural language. Describe a UI, get production-ready code with shadcn/ui and Tailwind. What makes it special: - Outputs clean, copy-paste-ready code - Uses shadcn/ui components (no proprietary stuff) - Iterates through conversation - Free tier for experimentation Built by Vercel, so the quality bar is high. Best for UI components rather than full apps. Surprisingly good at matching designs.
Bolt.new by StackBlitz is an AI-powered full-stack development environment that runs entirely in your browser. The workflow: - Describe your app in plain English - AI generates the full codebase - Preview runs instantly (WebContainers magic) - Edit with AI assistance, deploy when ready No local setup. No waiting for builds. It's the fastest path from idea to deployed app right now.
Lovable (formerly GPT Engineer) transforms prompts into full applications. It's moved from open-source experiment to serious product. What you get: - Full-stack app generation from descriptions - Supabase integration for backend - GitHub sync for version control - One-click deployment Competing directly with Bolt and Replit Agent. Strong on React/Next.js apps. Good for people who want more structure than vibe coding.
Codeium is the free alternative to GitHub Copilot. Same basic functionality, zero monthly cost. What you get for free: - Inline code completions - Chat interface in IDE - Support for 70+ languages - VS Code, JetBrains, Vim, and more The business model: enterprise sales subsidize free individual use. If you're cost-conscious or just starting out, Codeium delivers genuine value at $0.
free
Tabnine has been doing AI code completion since 2019 - before it was cool. It's enterprise-focused with strong privacy options. Key selling points: - Runs fully on-device (no code leaves your machine) - Learns from your codebase patterns - SOC 2 Type II certified - Self-hosted options for enterprises For companies worried about code privacy, Tabnine is the safest bet. Not the flashiest, but battle-tested and secure.
AgentGPT lets you spin up autonomous agents directly in the browser. No setup, no installation - just a goal and go. How it works: - Enter a goal in natural language - Agent breaks it into tasks - Executes and iterates autonomously - Shows its thinking and progress Open source and self-hostable. Great for experimenting with autonomous agents without commitment. Powered by the reworkd team.
Phidata is a framework for building production-ready AI assistants. Think LangChain but more opinionated and batteries-included. Core features: - Built-in memory (conversations persist) - Knowledge bases (RAG out of the box) - Tool use (web search, APIs, code execution) - Structured outputs that actually work Less flexible than LangChain, more productive for common use cases. Python-first, actively maintained, strong documentation.
