clauderon

Run AI coding agents in isolated, secure environments with zero-credential architectures

Why clauderon?

Zero-Credential Security

AI agents run with placeholder credentials. Real tokens are injected by the proxy at request time, keeping secrets out of agent memory.

Multiple Backends

Run sessions in Docker containers or Zellij terminal sessions. Choose the isolation level that fits your workflow.

TUI Interface

Full terminal user interface for creating, managing, and interacting with agent sessions.

Git Integration

Automatic git configuration injection. Sessions inherit your git identity for seamless version control.

Quick Install

# Download the latest release
curl -fsSL https://github.com/shepherdjerred/monorepo/releases/latest/download/clauderon-linux-x86_64 -o clauderon
chmod +x clauderon
sudo mv clauderon /usr/local/bin/

# Start a session
clauderon

Architecture

clauderon uses an HTTP proxy with TLS interception to securely manage credentials:

Session Creation: Create isolated environments (Docker/Zellij)
Proxy Configuration: Sessions route traffic through clauderon’s proxy
Credential Injection: OAuth tokens injected at request time
Request Filtering: Malicious requests blocked before reaching APIs

This architecture ensures AI agents never have direct access to your credentials.