ccp

Relay messages between users and Claude Code without modification.

27|1|Updated Mar 19, 2026
One-click install
npx skills add https://github.com/CorellisOrg/Corellis --skill ccp-corellisorg
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ccp
Source: https://github.com/CorellisOrg/Corellis/tree/main/templates/skills/ccp
Command: npx skills add https://github.com/CorellisOrg/Corellis --skill ccp-corellisorg

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides a transparent relay between a user and Claude Code, forwarding messages exactly as received without interpretation or modification.

Core Features & Use Cases

  • Forward messages between the user and Claude Code in real time.
  • Maintain a persistent session across multi-turn chats with a managed workspace.
  • Enforce non-interpreting behavior and safe, deterministic output delivery.

Quick Start

Activate by saying "ccp" or "ccp start" to begin a direct CC session, and then send messages to be relayed.

Frequently Asked Questions about ccp

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I relay messages directly to Claude Code without modification?

A direct relay forwards messages to Claude Code exactly as received without modification or commentary. It operates in direct CC sessions using -p and --resume mode, enforcing non-interpreting behavior for deterministic output.

How do I maintain a persistent session for multi-turn Claude Code chats?

Maintaining a persistent session for multi-turn Claude Code chats requires a managed workspace using --resume mode. This setup preserves context across multiple turns, forwarding CC output as-is throughout the conversation.

Can I select a specific workspace when starting a direct Claude Code connection?

Yes, optional workspace selection is supported when starting a direct Claude Code connection. You can activate a session and specify a workspace to isolate multi-turn chats and manage persistent contexts deterministically.

Why does my Claude Code relay add unwanted commentary to the output?

Unwanted commentary appears when the relay lacks strict non-interpretation enforcement. A pure relay forwards Claude Code output as-is with no modification or commentary, preventing any additional interpretation from altering the original output.

How does a pure relay handle timeouts and errors in Claude Code sessions?

A pure relay handles timeouts and errors gracefully by catching exceptions during direct Claude Code sessions. It ensures safe, deterministic output delivery without crashing the persistent multi-turn chat session.

What is the best way to forward Claude Code output as-is without interpretation?

The best way to forward Claude Code output as-is without interpretation is using a pure relay mechanism. It enforces non-interpreting behavior, forwarding messages exactly as received while managing multi-turn sessions in a persistent workspace.