Metadata-Version: 2.4
Name: specter-agent
Version: 0.1.0
Summary: The Agent That Kills All Agents — Self-hosted local AI agent
Author: Nabil Karem
License: MIT
Project-URL: Homepage, https://github.com/Nabilkarem911/SPECTER
Project-URL: Repository, https://github.com/Nabilkarem911/SPECTER
Keywords: ai,agent,cli,local,llm
Classifier: Development Status :: 3 - Alpha
Classifier: Environment :: Console
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.10
Description-Content-Type: text/markdown
Requires-Dist: pyyaml>=6.0
Requires-Dist: httpx>=0.27
Dynamic: requires-python

# 👻 SPECTER

**The open-source AI agent.**  
Like OpenClaw — but you own it 100%. No SaaS, no vendor lock-in, no one else touching your keys.

Run it on your VPS. On your laptop. Inside a container. Wherever you want.

SPECTER manages your server, browses the web, runs sub-agents, reads/writes files, and does anything you ask.

---

## Quick Start

### Option A: pip (laptop / local)

```bash
pip install specter
specter
```

> First time? It'll ask for your LLM API key. That's it.

Or start the Telegram bot:

```bash
specter-telegram
```

### Option B: Docker (VPS / server)

```bash
mkdir specter && cd specter
wget https://raw.githubusercontent.com/Nabilkarem911/SPECTER/main/docker-compose.yml
wget https://raw.githubusercontent.com/Nabilkarem911/SPECTER/main/.env.example -O .env

# Edit .env — add your keys
nano .env

docker compose up -d
```

---

## What SPECTER Can Do

| What | How |
|------|-----|
| Chat via terminal | `specter` |
| Chat via Telegram | `specter-telegram` |
| Run commands on your server | Built-in terminal tool |
| Browse the web | Web search & fetch |
| Read/write files | File tool |
| Run sub-agents | Agent Pool (for complex tasks) |
| Remember conversations | ChromaDB + SQLite memory |
| Learn from mistakes | Learning Engine |
| Use any LLM | DeepSeek, OpenAI, Claude, Gemini, Ollama, Groq... |

Your API keys stay on **your** machine. We never see them.

---

## Architecture

```
You (CLI / Telegram / API)
  └─ SPECTER Agent
       ├─ LLM Router → Any provider (DeepSeek, OpenAI, Claude...)
       ├─ Memory     → ChromaDB + SQLite
       ├─ Tools      → Terminal, File, Web, Browser
       ├─ Agent Pool → Sub-agents for complex tasks
       └─ Learning   → Improves over time
```

---

## Requirements

- Python 3.10+
- An LLM API key (DeepSeek is cheap: ~$0.27 per million tokens)
- (Optional) Docker for server deployment

---

## License

MIT — do whatever you want. This is yours.
