# Creed Space

> Open public-good infrastructure for AI safety, portable policy, and accountable deployment.

## Overview

Creed Space is open, nonprofit-built infrastructure for practical AI safety and accountability. Before an AI acts, Creed Space checks the constitution you set; every decision through the Gateway is signed into a verifiable receipt you can inspect. Free to inspect, run, and self-host, with paid hosted tiers for scale and teams.

Your AI has permissions. Does it have principles?

## Products

- **Guardian**: a free local safety layer for AI outputs and agent actions. Evaluates against the policies you write, on your hardware, with the provider you choose. Every decision is inspectable. Apache-2.0. Available as a desktop app and a Python package.
- **VCP (Value Context Protocol)**: a portable, open standard for encoding values context that travels with you across AI services. Demo and docs at valuecontextprotocol.org.
- **Fleet**: agent-governance product — audit trails and policy enforcement for autonomous agents.
- **Safety Stack**: the hosted API for constitutional safety evaluation and policy adjudication (PDP).
- **Safety Research**: published measurement and alignment research.

## Important Routes

- / - Home
- /guardian - Guardian (local safety layer)
- /vcp - Value Context Protocol
- /mission - Mission and approach
- /blog - Writing and updates
- /docs - Documentation
- /privacy - Privacy policy
- /terms - Terms of service

## API

The Safety Stack API is documented at https://docs.creed.space

## Technology

- Frontend: Svelte 5, TypeScript
- Backend: Python, FastAPI
- Database: PostgreSQL
- Caching: Redis

## Contact

Website: https://creed.space