botron

Files (1)

• SKILL.md

SKILL.md (first lines)

# Botron — Autonomous Red Team Agent

Botron is a Docker-based autonomous red team agent with 17 specialist agents orchestrated through LangGraph. It routes all LLM calls through a LiteLLM proxy supporting **10+ AI providers** with automatic failover.

Fork of [Decepticon](https://github.com/PurpleAILAB/Decepticon) by PurpleAILAB, enhanced with multi-provider support and stripped of proprietary Claude Code OAuth dependencies.

## When to Use

- Setting up Botron for the first time
- Running the Metasploitable 2 demo
- Managing Docker Compose services (start, stop, logs, health)
- Running the Python test suite
- Configuring LLM providers
- Debugging service startup issues

## Quick Start

```bash
# Clone and enter (or use the Hermes tap: hermes skills tap add dazeb/botron)
cd ~/projects && git clone https://github.com/dazeb/botron.git
cd botron

# Create .env from template
cp clients/launcher/internal/config/env.example .env
```

Edit `.env` — set at least one API key (`ANTHROPIC_API_KEY`, `OPENAI_API_KEY`, etc.) and add these defaults:
```
LITELLM_MASTER_KEY=sk-botron-master
LITELLM_SALT_KEY=sk-botron-salt
POSTGRES_PASSWORD=botron
NEO4J_PASSWORD=botron-graph
BOTRON_MODEL_PROFILE=eco
```

```bash
# Start all services
make dev
# or: docker compose up -d --build

# Open CLI in separate terminal
make cli
```

## Service Management

All commands from `~/projects/botron/`:

```bash
# Start services with hot-reload
make dev

# Start services (no watch mode)
docker compose up -d --build

# Check status
docker compose ps
make status

# Health checks
make health

# View logs
docker compose logs -f langgraph
docker compose logs -f litellm

# Stop (keep data)
docker compose down

# Stop + remove all volumes (full reset)
make clean
```

## Run the Demo

```bash
cd ~/projects/botron
make demo
```

Launches Metasploitable 2, loads a pre-built engagement, and runs the full kill chain autonomously: port scan → vsftpd exploit → Sliver C2 implant → credential harvesting → internal recon.

## Run Tests

```bash
cd ~/projects/botron

# Python tests (local venv)
python3 -m venv .venv && .venv/bin/pip install pytest httpx "langchain-core>=0.3.0" "langchain-openai>=0.3.0" pydantic pydantic-settings pyyaml
.venv/bin/python -m pytest tests/unit/llm/ tests/unit/core/ -v

# Go launcher tests
cd clients/launcher && go test ./...
```

For full test suite (including Docker-dependent tests), run inside the langgraph container after installing pytest there.

## Architecture

```
botron/
├── botron/              # Python package — agents, LLM, tools, middleware
│   ├── llm/             # Model factory, router, model definitions
│   ├── agents/          # 17 specialist agents + prompts
│   ├── tools/           # Security tools (web, reversing, cloud, AD, research)
│   └── core/            # Config, logging, engagement loop
├── config/
│   └── litellm.yaml     # LiteLLM proxy model routing (10+ providers)
├── containers/          # Dockerfiles for each service
├── clients/
│   ├── launcher/        # Go CLI (onboard wizard, launcher)
│   └── web/             # Next.js dashboard
└── langgraph.json       # LangGraph graph definitions (17 agents)
```

### Docker Compose Services

| Service | Port | Purpose |
|---------|------|---------|

From dazeb/botron, Apache-2.0.

More from dazeb/botron