The zero-trust execution environment for autonomous agent skills.

VaultSkill packages SKILL.md with embedded cryptographic signatures and explicit permission manifests. An agent cannot execute a VaultSkill without passing a decentralized policy check, completely preventing rogue actions such as unauthorized data exposure or malicious file execution.

Install  ·  Quick Start  ·  Features  ·  Architecture  ·  Demo

---

Installation

pip install vaultskill

Quick Start

vaultskill --help

Architecture

vaultskill/
├── pyproject.toml
├── README.md
├── src/
│   └── vaultskill/
│       ├── __init__.py
│       └── cli.py
├── tests/
│   └── test_vaultskill.py
└── AGENTS.md

Demo

Coming soon

Development

git clone https://github.com/avasis-ai/vaultskill
cd vaultskill
pip install -e .
pytest tests/ -v

Links

• Repository: https://github.com/avasis-ai/vaultskill
• PyPI: https://pypi.org/project/vaultskill
• Issues: https://github.com/avasis-ai/vaultskill/issues

---

Part of the AVASIS AI open-source ecosystem