Self-hosted workflow automation with AI agents.
Weave your workflows together.

---

Weavr connects AI agents with your developer tools—GitHub, Linear, Notion, Slack, databases, CI/CD, and more. Think Zapier/n8n, but with native AI agent support and full data sovereignty.

✨ Features

• 🏠 Self-hosted — Your workflows, your data, your infrastructure
• 🤖 AI Agents — Autonomous agents with web search, file access, shell commands, and multi-step reasoning
• 🔌 Plugin ecosystem — Built-in integrations for GitHub, Slack, Discord, Telegram, Linear, Notion, and more
• 🧩 MCP Support — Extend AI agents with any Model Context Protocol server
• 📡 Real-time — WebSocket-powered event streaming
• 🎯 DAG execution — Parallel steps, conditional logic, retries, error handling
• 🖥️ CLI + Web UI — Terminal power users and visual builders welcome
• 💬 Messaging — Built-in WhatsApp, Telegram, Discord, and iMessage support
• 📧 Email & Calendar — Send emails and manage calendar events

🚀 Quick Start

# Install
npm install -g @openweavr/weavr

# Setup
weavr onboard

# Start the gateway
weavr serve

# Create your first workflow
weavr create

📖 Example Workflows

AI Research Agent

name: daily-market-research
description: AI agent researches market data and sends a report

trigger:
  type: cron.schedule
  with:
    expression: "0 9 * * *"  # Every day at 9am

steps:
  - id: research
    action: ai.agent
    with:
      tools: "web_search,web_fetch"
      task: |
        Research current market conditions:
        - Gold and silver prices vs USD
        - Top investment opportunities
        - Key financial news from reliable sources
        Compile into an easy-to-read report.

  - id: notify
    action: whatsapp.send
    needs: [research]
    with:
      to: "+1234567890"
      text: "{{ steps.research.result }}"

GitHub to Slack Notifications

name: bug-to-slack
description: Notify Slack when GitHub issues are labeled 'bug'

trigger:
  type: github.issue.labeled
  with:
    label: bug

steps:
  - id: notify
    action: slack.post
    with:
      channel: "#bugs"
      message: "🐛 New bug: {{ trigger.issue.title }}\n{{ trigger.issue.html_url }}"

See more examples in the examples/ directory.

🤖 AI Agent Tools

AI agents (ai.agent action) can use these built-in tools:

Tool	Description
web_search	Search the web (requires Brave or Tavily API key)
web_fetch	Fetch and extract content from URLs
http_request	Make HTTP/API requests
read_file	Read file contents
write_file	Create or modify files
list_directory	List files in a directory
shell_exec	Execute shell commands

Specify tools in your workflow:

- id: research
  action: ai.agent
  with:
    tools: "web_search,web_fetch"  # or "all" for all tools
    task: "Research the latest news on AI"

You can also extend agents with MCP servers for custom tool access.

🔌 Built-in Integrations

Plugin	Actions	Triggers
ai	agent, complete, summarize, classify, sentiment	—
github	create_issue, comment, get_issue, add_label	push, pull_request, issue, release, etc.
slack	post, reply	message, reaction
discord	send, reply	message
telegram	send, reply	message
whatsapp	send	—
imessage	send	—
linear	create_issue, update_issue	—
notion	create_page, update_page, query	—
email	send	—
calendar	create_event, list_events	—
http	request, fetch	webhook
cron	—	schedule
shell	exec	—
filesystem	read, write, list	—

🛠️ CLI Commands

# Setup & Diagnostics
weavr onboard          # Interactive setup wizard
weavr doctor           # Diagnose configuration issues

# Server
weavr serve            # Start gateway server
weavr serve --port 3847

# Workflows
weavr list             # List all workflows
weavr run <name>       # Run a workflow manually
weavr create           # Create new workflow (interactive)

# AI (coming soon)
weavr ask "When PR is merged, deploy to staging"

🏗️ Architecture

┌─────────────────────────────────────────────────────────────┐
│                      Gateway Server                          │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────────────┐  │
│  │  HTTP API   │  │  WebSocket  │  │  Webhook Receiver   │  │
│  └─────────────┘  └─────────────┘  └─────────────────────┘  │
└─────────────────────────────────────────────────────────────┘
                              │
              ┌───────────────┼───────────────┐
              ▼               ▼               ▼
      ┌─────────────┐ ┌─────────────┐ ┌─────────────┐
      │   Engine    │ │   Plugins   │ │  AI Agent   │
      │ (DAG exec)  │ │  (GitHub,   │ │  (Natural   │
      │             │ │  Slack...)  │ │  language)  │
      └─────────────┘ └─────────────┘ └─────────────┘

⚙️ Configuration

AI Provider

Configure your AI provider during onboarding or in Settings:

• Anthropic — Claude models (default: claude-sonnet-4-20250514)
• OpenAI — GPT models (default: gpt-4o). Supports OAuth sign-in or API key.
• Ollama — Local models (default: llama3.2). Any Ollama-supported model works.

You can specify any model your provider supports in ~/.weavr/config.yaml:

provider: anthropic  # or openai, ollama
model: claude-sonnet-4-20250514  # any model ID supported by your provider

Web Search (for AI Agents)

AI agents need a search API to browse the web. Get a free Brave Search API key:

1. Sign up at brave.com/search/api
2. Choose "Data for Search" plan (2,000 free queries/month)
3. Add your API key in Settings or set BRAVE_API_KEY environment variable

Alternatively, you can use Tavily Search API by setting TAVILY_API_KEY.

Environment Variables

Variable	Description
ANTHROPIC_API_KEY	Anthropic API key for Claude models
OPENAI_API_KEY	OpenAI API key for GPT models
BRAVE_API_KEY	Brave Search API key (for AI agent web search)
TAVILY_API_KEY	Tavily Search API key (alternative to Brave)
GITHUB_TOKEN	GitHub personal access token (for GitHub triggers/actions)
SLACK_BOT_TOKEN	Slack bot token (for Slack integration)
DISCORD_BOT_TOKEN	Discord bot token (for Discord integration)
TELEGRAM_BOT_TOKEN	Telegram bot token (for Telegram integration)
LINEAR_API_KEY	Linear API key (for Linear integration)

📁 Project Structure

~/.weavr/
├── config.yaml        # Global configuration
├── workflows/         # Your workflow definitions
├── plugins/           # Installed plugins
├── logs/              # Execution logs
└── scheduler.db       # Scheduler queue/state (configurable)

You can override the scheduler database location in ~/.weavr/config.yaml:

scheduler:
  dbPath: /path/to/scheduler.db

🔌 Creating Plugins

import { definePlugin, defineAction } from 'weavr/plugins';

export default definePlugin({
  name: 'my-plugin',
  version: '1.0.0',

  actions: [
    defineAction({
      name: 'greet',
      execute: async (ctx) => {
        const name = ctx.config.name as string;
        ctx.log(`Hello, ${name}!`);
        return { greeted: name };
      },
    }),
  ],
});

📚 Documentation

Full documentation is available at openweavr.github.io/Openweavr.

• Getting Started
• Writing Workflows
• AI Agents
• Integrations
• CLI Reference

🤝 Contributing

We welcome contributions! AI-assisted PRs are explicitly encouraged.

See CONTRIBUTING.md for guidelines.

📜 License

MIT © Weavr Contributors