Claude Code Integration Guide

Overview

Claude Code is a terminal-first coding agent that can read files, edit repositories, and execute commands. Cup’n’String treats it as a high-value governance target.

Support level

Active Proxy & Shielding

What Cup’n’String detects

• Claude Code processes
• Shell/session context
• Workspace path
• Provider/API egress

What it governs

• Outbound model calls
• Shell command network behavior
• Sensitive file access patterns where visible
• Local service access

Recommended policies

• Require provider/API proxy for Claude Code
• Audit terminal-agent sessions
• Block unapproved localhost and metadata-service access
• Shield credentials and .env files

Setup outline

1. Ensure the Cup’n’String agent is active.
2. Initialize Claude Code in your terminal.
3. Configure the local proxy settings or wrapper scripts to intercept CLI commands and outbound model requests.

Verification

Confirm terminal commands executed by Claude Code show up in the audit logs, and try executing an unapproved command to verify blocking policy triggers.

Troubleshooting

If CLI outputs are not captured, check that shell wrapper functions are initialized in your .bashrc or .zshrc.

Known limitations

Terminal agents vary in what they expose unless routed through managed proxy/wrapper paths.