Skip to content

nwang783/tempo-sec

Repository files navigation

Tempo Sec

A cybersecurity agent for autonomous vulnerability discovery. Uses deliberate context management, human-like workflow phases, and sandboxed tool execution.

Python 3.11+ License


Quick Start

1. Clone and Install

git clone https://github.com/nwang783/tempo-sec.git
cd tempo-sec

# Create virtual environment
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install dependencies
pip install -e .

2. Configure

# Copy example config
cp config.example.toml config.toml

# Edit with your API keys
# Required: ANTHROPIC_API_KEY or OPENAI_API_KEY or GEMINI_API_KEY in .env
cp .env.example .env

3. Run the Security Analysis

Recommended: Use the Orchestrator (fully autonomous)

# Run the Orchestrator Agent - coordinates recon and testing automatically
tempo --agent orchestrator --target https://mysouschef.com

# With custom findings path and max iterations
tempo --agent orchestrator --target https://example.com \
    --findings ./findings/example.yaml \
    --max-iterations 15

Documentation

Document Description
MISSION.md Vision, guiding principles, philosophy
AGENTS.md AI coding assistant guidance
docs/ARCHITECTURE.md Component architecture, structure
docs/WORKFLOW.md Phase workflow, branching
docs/TOOLS.md Tool specifications
diagrams/ Architecture diagrams

Project Structure

tempo-sec/
├── tempo/
│   ├── agent/          # Agent orchestration, workflow
│   ├── context/        # State management, persistence
│   ├── prompts/        # LLM instruction templates
│   ├── tools/          # Shell, browser, meta tools
│   ├── visualization/  # Real-time UI server
│   └── infra/          # LLM client, container, protocols
├── docs/               # Detailed documentation
├── diagrams/           # Architecture diagrams
├── tests/              # Test suite
└── config.toml         # Configuration

Development

Run Tests

python -m pytest tests/ -v

Architecture Rules

Note to Ryan: This is my attempt to separate the main concerns of building an agentic systen. We can change this if you have a better idea.

CRITICAL: Context, Prompts, and Tools MUST NOT import from each other.
They may only import from infra/ or external packages.
Only Agent composes them together.

See AGENTS.md for complete development guidelines.


Contributing

Note to Ryan: This is in here in case we do open source.

Contributions welcome! Please:

  1. Fork the repository
  2. Create a feature branch
  3. Run tests before submitting PR
  4. Follow the architecture rules above

License

Apache 2.0 - See LICENSE for details.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors