I Love ❤️ Agents

AI Agents, ready to use. Open source. Community-built. Bring your own key.

---

Table of Contents

• What is iloveAgents?
• Why iloveAgents?
• Available Agents
• Supported Providers
• Battle Mode

• AI Workflow Builder (New)

• What you can do

• How sequential execution works
• Navigation
• Getting Started
• Prerequisites
• Run Locally
• Deploy Your Own
• How It Works
• Contributing
• Add a New Agent in 3 Steps
• Other Ways to Help
• Tech Stack
• Community
• License

---

What is iloveAgents?

iloveAgents is a clean, open source web platform where you can run AI agents directly in your browser.

No sign-up. No backend. No data collection. Just paste your API key and go.

Each agent is a focused tool that does one thing really well — summarize meeting notes, review code, generate SQL, write cold emails, and a lot more. The whole platform is config-driven, which means adding a new agent is as simple as adding one JavaScript object to a single file. No deep React knowledge needed.

---

Why iloveAgents?

• Your API key never leaves your browser. It goes directly to OpenAI, Anthropic, or Google. No middleman, no storage, no tracking.
• Zero setup. No .env files, no backend, no database. Clone and run in under a minute.
• Works with all major providers. OpenAI, Anthropic, and Google Gemini — switch between them at runtime.
• Built for contributors. Adding a new agent takes about 5 minutes and you don't need to know much React at all.

---

Available Agents

The complete list of agents has been moved to AGENTS.md for better organization and scalability.

---

Supported Providers

Provider	Logo	Models	Get Your Key
OpenAI		GPT-4o, GPT-4o-mini	platform.openai.com
Anthropic		Claude Opus, Claude Sonnet	console.anthropic.com
Google Gemini		Gemini 2.5 Flash	aistudio.google.com

You can switch providers on any agent at runtime from the dropdown. No restart needed.

---

Battle Mode

Pit three AI providers against each other head-to-head.

1. Pick any agent from the full registry
2. Enter your input once — same prompt goes to all three
3. GPT-4o vs Claude Sonnet vs Gemini Flash generate outputs simultaneously
4. You pick the winner

Battle Mode has its own dark, dramatic UI with color-coded provider columns (gold for OpenAI, purple for Anthropic, blue for Gemini). Each provider loads independently — if one fails, the other two still work. Access it from the "Enter Battle Mode" button on the homepage or navigate directly to /battle.

---

🔗 AI Workflow Builder (New)

Chain multiple agents together and automate your entire process in one run.

Workflows let you connect agents in sequence — the output of each agent automatically becomes the input for the next. Build once, run with any input.

What you can do

• Build a workflow — pick up to 5 agents, arrange them in order, give the workflow a title
• Run it in one click — paste your input once, every step runs automatically in sequence
• Watch it execute — each step shows its own live status: waiting → running → done / failed
• Chain any agents — research → summarize → write LinkedIn post → done
• Community library — browse and run workflows shared by other users
• Real-time counters — usage counts update live as others run the same workflows
• Share workflows — one-click URL copy on any workflow detail page

How sequential execution works

Your Input
    │
    ▼
┌─────────────┐
│  Agent  1   │  e.g. Research Agent
└──────┬──────┘
       │ output
    ▼
┌─────────────┐
│  Agent  2   │  e.g. PDF Summarizer
└──────┬──────┘
       │ output
    ▼
┌─────────────┐
│  Agent  3   │  e.g. LinkedIn Post Writer
└─────────────┘
       │
    Final Output

If any step fails, the workflow stops at that step and shows you exactly what went wrong — with a Retry button. On success, you can copy all outputs at once.

Navigation

Route	What it does
/workflows	Browse community workflow library
/workflows/build	Create and save a new workflow
/workflows/:id	View full details of a workflow
/workflows/:id/run	Run a workflow step-by-step

---

Getting Started

Prerequisites

• Node.js v18+
• npm or yarn

Run Locally

# Clone the repository
git clone https://github.com/AditthyaSS/iloveAgents.git
cd iloveAgents

# Install dependencies
npm install

# Start the dev server
npm run dev

Open http://localhost:5173 in your browser. No API provider keys are required in a .env file because they are entered at runtime and never stored anywhere.

However, local development requires a .env.local file for Supabase features like Workflows.

Create a .env.local file in the root directory:

VITE_SUPABASE_URL=your_supabase_url
VITE_SUPABASE_ANON_KEY=your_supabase_anon_key

Note: AI provider API keys are still entered at runtime and are never stored anywhere.

Deploy Your Own

1. Fork this repository
2. Import to Vercel or any static host
3. Deploy — zero configuration needed

The included vercel.json handles SPA routing automatically.

---

How It Works

src/
├── agents/
│   ├── definitions/          # Each agent in its own file (auto-collected)
│   │   ├── pdf-summarizer.js
│   │   ├── code-reviewer.js
│   │   └── ... more agent definitions
│   ├── categories.js         # CATEGORIES constant
│   └── registry.js           # Auto-collects all definitions via import.meta.glob
├── components/
│   ├── AgentRunner.jsx       # Generic agent execution UI
│   ├── AgentCard.jsx         # Agent card for the homepage grid
│   ├── ApiKeyBar.jsx         # Provider and API key input
│   ├── OutputRenderer.jsx    # Renders markdown/text/JSON output
│   ├── ScorecardOutput.jsx   # Visual scorecard for JSON output
│   ├── BattleNavbar.jsx      # Battle Mode navigation bar
│   ├── Navbar.jsx            # Top navigation
│   ├── Sidebar.jsx           # Agent sidebar
│   └── ...
├── hooks/
│   └── useWorkflows.js       # Workflow data operations (fetch, save, realtime)
├── lib/
│   ├── llmAdapter.js         # Unified API adapter for all providers
│   ├── supabase.js           # Realtime data client
│   └── useApiKey.js          # API key state management
├── pages/
│   ├── HomePage.jsx          # Landing page with agent grid
│   ├── AgentPage.jsx         # Individual agent page
│   ├── WorkflowLibrary.jsx   # 🆕 Public workflow library with live counters
│   ├── WorkflowBuilder.jsx   # 🆕 Drag-and-drop agent chain builder
│   ├── WorkflowDetail.jsx    # 🆕 Single workflow view with realtime stats
│   ├── WorkflowRunner.jsx    # 🆕 Sequential agent execution engine
│   ├── BattleModeLanding.jsx # Battle Mode entry page
│   ├── BattleModeSetup.jsx   # Battle configuration
│   ├── BattleModeArena.jsx   # Three-column battle arena
│   └── BattleModeWinner.jsx  # Winner announcement
└── main.jsx

1. Registry — Each agent is its own file in src/agents/definitions/. The registry auto-collects them via import.meta.glob — just drop a file in and it appears.
2. LLM Adapter — A single runAgent() function in llmAdapter.js handles all three providers through one unified interface.
3. Agent Runner — AgentRunner.jsx builds the input form from the config, constructs the prompt, and renders the response.
4. Battle Mode — BattleModeArena.jsx fires the same prompt to GPT-4o, Claude Sonnet, and Gemini Flash simultaneously and lets you pick the winner.
5. Workflow Builder — WorkflowRunner.jsx chains agents sequentially using the same runAgent() adapter — output of step N becomes input of step N+1, with per-step status cards and real-time usage counters.
6. No backend — Every API call goes directly from your browser to the provider. Nothing passes through our servers because there are no servers.

---

Contributing

iloveAgents is built by the community. Every contribution matters — whether it is a new agent, a bug fix, a UI improvement, or just fixing a typo.

Add a New Agent in 3 Steps

1. Create a new file in src/agents/definitions/ named your-agent-id.js:

export default {
  id: 'your-agent-id',
  name: 'Your Agent Name',
  description: 'One-line description.',
  category: 'Category',
  icon: 'IconName',               // from lucide.dev/icons
  provider: 'any',                // 'openai' | 'anthropic' | 'gemini' | 'any'
  defaultProvider: 'openai',
  model: 'gpt-4o',
  inputs: [
    {
      id: 'field_id',
      label: 'Field Label',
      type: 'textarea',           // text | textarea | code | select | multiselect
      placeholder: 'Hint text...',
      required: true,
    },
  ],
  systemPrompt: `Your system prompt here.`,
  outputType: 'markdown',         // markdown | text | json
}

The registry auto-collects it — no need to edit registry.js.

2. Run npm run dev and test your agent with a real API key.

3. Open a PR. That is it!

See CONTRIBUTING.md for the full guide.

Other Ways to Help

• 🐛 Fix a bug
• 🎨 Improve the UI or UX
• ♿ Improve accessibility
• 📝 Improve the docs
• 🧪 Add tests

---

Tech Stack

Technology	Purpose
React 18	Component framework
Vite 6	Build tool and dev server
Tailwind CSS 3	Styling
React Router 6	Client-side routing
Lucide React	Icons
react-markdown	Markdown rendering
react-syntax-highlighter	Code highlighting

---

Community

• Support — stuck on something? Drop a comment on your issue and I will reply within 24 hours
• Hall of Fame — every person who has contributed to iloveAgents
• Maintainers — who runs this project

---

👥 Contributors

Thanks to all contributors ❤️

---

License

Licensed under the MIT License — use it, fork it, build on it.

---

Built with ❤️ by @AditthyaSS and the open source community

⭐ Star this repo if you find it useful — it helps others discover it!

Live Demo · Report a Bug · Request an Agent · Contribute