Skip to content

Swimhack/strickland-agent-starter

BranchesTags

Repository files navigation

🤖 Strickland Agent Starter

Your personal AI agent - like JARVIS, but yours to own.

Runs on your own infrastructure. 100% private.

🎯 What Is This?

Strickland Agent Starter is a turnkey AI agent framework that gives you a personal assistant like JARVIS from Iron Man. It runs on your own server, connects to your accounts, and helps you with everything from emails to calendars to business automation.

Think of it as:

  • Your personal AI assistant that never sleeps
  • Proactive (checks things for you, reminds you, takes action)
  • Private (your data never leaves your server)
  • Extensible (add skills and integrations easily)

✨ Features

Core Capabilities

  • 🧠 Persistent Memory - Remembers conversations, decisions, context
  • 💬 Multi-Platform Messaging - Telegram, Discord, Signal, WhatsApp
  • 📧 Email Management - Read, send, organize (Gmail, Outlook, Zoho)
  • 📅 Calendar Integration - Track events, set reminders
  • 🌐 Web Research - Search, fetch content, monitor sites
  • 📝 File Management - Read, write, organize your documents
  • ⚡ Automation - Cron jobs, webhooks, background tasks

Advanced Features (New!)

  • 📞 AI Voice Calling - Make and receive calls with custom AI personas
  • 💼 Sales Outreach - Automated voice/email/SMS campaigns (optional)
  • 🎙️ Custom Voice Agents - Create AI agents with any personality
  • 📊 CRM Integration - Track leads and touchpoints
  • 🤖 Sub-Agents - Spawn background tasks with isolated context
  • 🔄 Product Maintainer - Automatically sync your improvements to GitHub (NEW!)

🔒 Security First

AI agents are powerful — and power requires responsibility. Recent security research has highlighted risks with self-hosted AI assistants. This starter addresses those concerns head-on.

The Concerns (and Our Answers)

Concern Risk This Starter's Approach
Exposed admin ports Hundreds of instances found publicly accessible Our config defaults to localhost binding only. Never expose port 18860 to the internet without VPN/tunnel + auth.
Plaintext secrets API keys in Markdown/JSON files Use environment variables or encrypted secret stores. Our .gitignore excludes all sensitive files. Never commit config.json.
Reverse proxy misconfig Internet traffic treated as localhost (auto-auth bypass) Enable gateway.authToken immediately. Don't rely on IP-based trust.
Skill library poisoning Malicious skills on ClawdHub can execute code We don't auto-install skills. Review any skill before adding. Pin versions.
Prompt injection Malicious messages via WhatsApp/email trigger unintended actions Configure exec.security: "allowlist" to restrict commands. Use exec.ask: "always" for destructive ops.
Infostealer targeting Malware hunting AI agent directories Run your agent in a dedicated VM or container. Don't run on your primary workstation with banking sessions.

Security Checklist

Before going live, verify:

# ✅ Auth token is set (not empty)
grep -q '"authToken":' config/openclaw.json && echo "✅ Auth configured" || echo "❌ Set authToken!"

# ✅ Gateway binds to localhost only
grep -q '"host": "127.0.0.1"' config/openclaw.json && echo "✅ Localhost only" || echo "⚠️ Exposed to internet!"

# ✅ Exec restricted to allowlist
grep -q '"security": "allowlist"' config/openclaw.json && echo "✅ Exec restricted" || echo "⚠️ Unrestricted exec!"

# ✅ No secrets in git
git status --porcelain | grep -v '^\?\?' | grep -E '\.(json|env)$' && echo "⚠️ WARNING: secrets may be staged" || echo "✅ Clean"

Recommended Security Posture

  1. Never expose to public internet directly - Use Tailscale/WireGuard/SSH tunnel
  2. Enable authentication - Set strong authToken in config
  3. Use environment variables - Keep API keys out of config files
  4. Review skills before installation - Treat like installing software
  5. Run in dedicated VM/container - Isolate from personal data
  6. Keep updated - Security patches matter
  7. Monitor logs - Watch for suspicious activity

Bottom line: This is a powerful tool. Use it responsibly. If you're deploying for a team or organization, consult a security professional.

🚀 Quick Start

Prerequisites

  • Ubuntu/Debian VPS (4GB+ RAM recommended)
  • Node.js 18+
  • Basic command line knowledge

Installation

# Clone the starter
git clone https://github.com/stricklandtechnology/agent-starter.git
cd agent-starter

# Run setup script
./setup.sh

# Follow prompts to configure:
# - Your name and timezone
# - Messaging platform (Telegram/Discord/etc.)
# - AI provider (OpenAI, Anthropic, local)
# - Optional integrations (email, calendar, voice)

That's it! Your agent is running.

📦 What's Included

Core Files

agent-starter/
├── workspace/              # Your agent's workspace
│   ├── AGENTS.md          # How your agent operates
│   ├── SOUL.md            # Your agent's personality
│   ├── USER.md            # About you
│   ├── MEMORY.md          # Long-term memory
│   ├── HEARTBEAT.md       # Proactive tasks
│   └── memory/            # Daily logs
├── skills/                # Extensible skills
│   ├── voice-calling/     # AI voice agent skill
│   ├── email-outreach/    # Email campaign skill
│   ├── weather/           # Weather reports
│   └── tmux/              # Terminal control
├── config/
│   └── openclaw.json      # Agent configuration
└── setup.sh               # One-click setup

Pre-Built Skills

  • Voice Calling (Vapi.ai integration)
  • Email Campaigns (SendGrid/Postmark)
  • SMS Messaging (Twilio)
  • Weather (wttr.in)
  • Terminal Control (tmux)
  • Web Research (Brave Search)

🎨 Customization

Change Your Agent's Personality

Edit workspace/SOUL.md:

# SOUL.md - Who You Are

Be direct and efficient. Skip the fluff.
When you don't know something, say so.
Be proactive - anticipate what I need.

Add Your Context

Edit workspace/USER.md:

# USER.md - About Your Human

- **Name:** Your Name
- **Timezone:** America/Chicago
- **Work:** What you do
- **Preferences:** How you like to communicate

Configure Integrations

Edit config/openclaw.json:

{
  "model": "anthropic/claude-sonnet-4",
  "skills": ["voice-calling", "email-outreach"],
  "integrations": {
    "telegram": { "enabled": true, "token": "YOUR_TOKEN" },
    "vapi": { "enabled": true, "apiKey": "YOUR_KEY" }
  }
}

🔌 Optional Add-Ons

Voice Calling Setup

  1. Get Vapi.ai API key: https://vapi.ai
  2. Add to config:
{
  "integrations": {
    "vapi": {
      "enabled": true,
      "apiKey": "YOUR_VAPI_KEY"
    }
  }
}
  1. Tell your agent: "Create a voice agent named Sarah"

Email Campaigns

  1. Get SendGrid API key: https://sendgrid.com
  2. Add to config:
{
  "integrations": {
    "sendgrid": {
      "enabled": true,
      "apiKey": "YOUR_KEY",
      "fromEmail": "you@yourdomain.com"
    }
  }
}
  1. Tell your agent: "Send an email campaign to my lead list"

Business Automation (Outreach Engine)

Want full multi-tenant SaaS capabilities?

Option 1: DIY (for developers)

  • Deploy Strickland Outreach Engine yourself
  • Connect via API
  • Full control, but more setup

Option 2: Hosted (coming soon)

  • We host and manage everything
  • $99/month add-on to your agent
  • Zero setup, just enable and go

💰 Pricing

Strickland Agent Starter:

  • Personal Use: FREE (open source, self-hosted)
  • Commercial License: $499 one-time (if reselling/white-label)

Optional Add-Ons:

  • Hosted Outreach Engine: $99/month (multi-tenant sales automation)
  • Priority Support: $49/month
  • Custom Skills: $299 one-time per skill

🛡️ Privacy & Security

Your Data, Your Server:

  • ✅ Runs on your own VPS (we never see your data)
  • ✅ API keys stored locally (encrypted)
  • ✅ Open source (audit the code yourself)
  • ✅ No telemetry, no tracking, no backdoors

API Keys You Control:

  • Vapi.ai (voice calling)
  • SendGrid/Postmark (email)
  • Twilio (SMS)
  • Anthropic/OpenAI (AI models)

We never store your API keys or personal data.

📖 Documentation

  • Getting Started: docs/quickstart.md
  • Configuration: docs/configuration.md
  • Skills Guide: docs/skills.md
  • API Reference: docs/api.md
  • Troubleshooting: docs/troubleshooting.md

🤝 Community

🚀 Use Cases

Personal Assistant

  • Check emails, calendar, weather
  • Set reminders and take notes
  • Research topics and summarize content
  • Automate daily tasks

Business Automation

  • Make AI voice calls to leads
  • Send personalized email campaigns
  • Track conversations in CRM
  • Generate reports and insights

Developer Productivity

  • Control terminal sessions
  • Deploy code automatically
  • Monitor servers and services
  • Run tests and CI/CD pipelines

Content Creation

  • Research and compile information
  • Draft emails and documents
  • Generate social media posts
  • Summarize articles and videos

🔧 Advanced

Create Custom Skills

cd skills
mkdir my-skill
cd my-skill
touch SKILL.md script.js

SKILL.md:

# My Custom Skill

**Description:** What this skill does

**Usage:** How to invoke it

**Requirements:** APIs or dependencies needed

script.js:

#!/usr/bin/env node
// Your skill logic here
console.log('Skill executed!');

Deploy to Production

# Install as systemd service
sudo cp agent-starter.service /etc/systemd/system/
sudo systemctl enable agent-starter
sudo systemctl start agent-starter

# Monitor logs
sudo journalctl -u agent-starter -f

Scale with Docker

# Build container
docker build -t strickland-agent .

# Run
docker run -d \
  -v ./workspace:/app/workspace \
  -v ./config:/app/config \
  --name my-agent \
  strickland-agent

🔄 Product Maintainer (Auto-Sync to GitHub)

Turn your production improvements into product features automatically.

The Product Maintainer is a dedicated sub-agent that monitors your daily work and automatically syncs publishable features to your public GitHub repository - with full data sanitization.

How It Works:

  1. You build a new feature during your daily work (e.g., email integration)
  2. Product Maintainer detects it that night (daily cron at 11 PM)
  3. Copies the code to your public repo
  4. Sanitizes personal data (API keys, phone numbers, addresses)
  5. Updates documentation automatically
  6. Commits and pushes to GitHub
  7. Notifies you: "🚀 Pushed email-inbox skill to GitHub (v2.0.2)"

Setup:

# 1. Copy the agent definition
cp -r agents/product-maintainer ~/.openclaw/workspace/agents/

# 2. Configure your GitHub credentials
export GITHUB_TOKEN="your_token_here"

# 3. Set up daily cron job
# Tell your agent: "Set up product maintainer to run daily at 11 PM"

What Gets Sanitized:

  • ❌ API keys (Vapi, OpenAI, Anthropic, Twilio, etc.)
  • ❌ Phone numbers and email addresses
  • ❌ Physical addresses
  • ❌ Customer names and business data
  • ✅ Replaces with environment variables and placeholders

Why This Matters:

  • Every internal improvement becomes a sellable feature
  • Zero manual effort to maintain open-source version
  • Community sees active development (builds trust)
  • Your production work creates value for thousands of users

Manual Trigger:

# Sanitize and push current work immediately
./scripts/sanitize-for-github.sh
cd your-public-repo
git add .
git commit -m "Add new feature"
git push

See agents/product-maintainer/AGENT.md for full documentation.

📊 Comparison

vs. ChatGPT:

  • ✅ Persistent memory
  • ✅ Proactive (not reactive)
  • ✅ Connects to your accounts
  • ✅ 100% private (your server)

vs. Custom AI Agent:

  • ✅ Turnkey (no coding required)
  • ✅ Pre-built skills included
  • ✅ Production-ready
  • ✅ Community support

vs. Zapier/n8n:

  • ✅ Conversational interface
  • ✅ AI-powered decision making
  • ✅ Multi-channel (not just workflows)
  • ✅ Learning and adaptation

🎯 Roadmap

v1.0 (Current):

  • ✅ Core agent framework
  • ✅ Multi-platform messaging
  • ✅ Memory and context
  • ✅ Basic skills (weather, web, files)

v1.5 (Next Month):

  • ⏳ Voice calling skill (Vapi integration)
  • ⏳ Email campaign skill
  • ⏳ CRM integration skill
  • ⏳ One-click deploy to Fly.io

v2.0 (Q2 2026):

  • 📋 Visual skill builder (no-code)
  • 📋 Mobile app (iOS/Android)
  • 📋 Team collaboration (multi-user)
  • 📋 Marketplace (buy/sell skills)

📝 License

MIT License (for personal use)

Commercial License Required for:

  • Reselling as a service
  • White-labeling
  • Including in paid products

Contact: licensing@stricklandtechnology.net

🙏 Credits

Built by Strickland Technology
Powered by OpenClaw
Inspired by JARVIS (Iron Man)

Turn your server into your personal AI assistant. Get started in 5 minutes.

curl -sSL https://get.stricklandagent.com | bash

🚀 Let's go.

About

No description, website, or topics provided.

Resources

Readme

License

MIT license
Activity

Stars

2 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

3 tags

Packages

 
 
 

Contributors

Languages