功能描述

Claude Code relay via tmux. Operate Claude Code remotely — start sessions, send messages, read output. Use when the user wants to interact with Claude Code f...

使用说明 (SKILL.md)

cc — Claude Code Relay

Name: Cc
Author: artwalker

Operate Claude Code remotely from any OpenClaw channel via tmux.

Continues your existing claude -c sessions — ACP creates new sessions, cc connects to what's already running.

Language: Always reply in the same language the user uses.

Script

{baseDir}/scripts/cc.sh \x3Ccommand> [args...]

Commands

Command	Action
`/cc on \x3Cproject>`	Start session (`claude -c` in project dir)
`/cc off [project]`	Stop session
`/cc ?`	Check Claude Code status (running/idle/dead)
`/cc tail [project] [lines]`	Show recent output
`/cc projects`	List available projects
`/cc status`	List active sessions
`/cc config root \x3Cpath>`	Set project root directory
`/cc`	Show help + project list
`/cc \x3Cmessage>`	Send message to Claude Code

Relay Mode (CRITICAL)

After /cc on \x3Cproject>, you enter relay mode:

ALL user messages are forwarded to Claude Code — NEVER answer yourself
Only messages NOT forwarded: /cc off, /cc ?, /cc tail, /cc status, /cc projects, /cc config
Relay mode ends on /cc off

You are a transparent pipe. Never interpret, analyze, or answer the user's question yourself.

Starting a Session

Run: scripts/cc.sh on \x3Cproject>

Report to user:

✅ Claude Code session started for \x3Cproject>
Your messages will now be sent directly to Claude Code.
Send /cc off to exit relay mode.

Enter relay mode

Sending Messages (relay flow)

When user sends a message in relay mode:

Immediately reply: ⏳ (so user knows message was received)
Forward: scripts/cc.sh \x3Cproject> "\x3Cuser's message>"
The script returns the incremental output — only content from this reply, not history
Return output to user (see Output Formatting below)

Stopping a Session

Run: scripts/cc.sh off \x3Cproject>

Report to user:

Session ended. You're back to normal chat.

Status Check (`/cc ?`)

Run: scripts/cc.sh check \x3Cproject>

Report result to user:

"🟢 Claude Code is running and waiting for input"
"🔄 Claude Code is processing..."
"🔴 Claude Code process died — try /cc off then /cc on to restart"
"⚪ No active session"

`/cc` (no arguments)

Run scripts/cc.sh projects. Show brief help + project list.

If there's a last-used project (marked with ★), show it first. Keep the response short — just names, no paths.

First-time Setup

When scripts/cc.sh projects outputs SETUP_NEEDED (exit 100):

Check: which tmux and which claude — report if missing
Ask user: "Where are your projects? (e.g., ~/projects)"
Run: scripts/cc.sh config root \x3Ctheir-answer>
List projects to confirm

Output Formatting

If output ≤ 4000 characters: wrap in one code block and send.

If output > 4000 characters: send a summary of the key output (first meaningful paragraph + last 10 lines), then add: "Full output: send /cc tail to see more"

Always: strip ANSI escape codes (the script handles this automatically).

Requirements

tmux installed
claude CLI installed (npm i -g @anthropic-ai/claude-code)
Any OpenClaw channel (Telegram, Discord, CLI, etc.)

安全使用建议

This skill is coherent with its stated purpose (it runs a `claude` process inside tmux and forwards messages), but installing it will: (1) cause the agent to forward all messages in 'relay mode' to your local `claude` CLI (and thus to Anthropic), which can leak sensitive information; (2) create and persist logs under ~/.local/state/cc/logs containing the conversation output; and (3) rely on your local `claude` CLI being configured with an API key (the skill does not declare or manage that key). Before installing: review scripts/cc.sh yourself, ensure you trust the Anthropic/claude CLI configuration, consider whether you want incremental logs written to disk (and rotate/secure them), and be cautious using relay mode with any sensitive data. If you need, I can point out the exact lines in the script that write logs and read optional env vars, or suggest mitigations (disable logging, change log path, require explicit confirmation before entering relay mode).

功能分析

Type: OpenClaw Skill Name: cc Version: 2.1.0 The skill provides a remote relay to the Claude Code CLI tool via tmux, which effectively grants the user remote command execution capabilities on the host. The SKILL.md file contains explicit instructions for the AI agent to act as a 'transparent pipe' and 'never interpret, analyze, or answer,' which is a prompt-injection technique designed to bypass the agent's safety filters and reasoning. While the bash script (scripts/cc.sh) appears to be a legitimate implementation of this relay functionality and lacks overt evidence of intentional malware or data exfiltration, the combination of remote shell access and instructions to disable agent-side oversight represents a high-risk security profile.

能力评估

✓ Purpose & Capability

Name/description (Claude Code relay via tmux) aligns with the included script and instructions: the skill starts tmux sessions running `claude -c`, sends user messages into the session, and returns incremental output. The declared required binaries (tmux, claude) are appropriate.

⚠ Instruction Scope

Runtime instructions put the agent into a persistent 'relay mode' that forwards ALL user messages to the Claude Code process and instructs the agent to never answer itself. This is consistent with the stated purpose but is high-risk for accidental data leakage: any sensitive user content will be forwarded to the Anthropic endpoint used by the `claude` CLI. The SKILL.md does not explicitly warn that forwarded content is sent to an external service nor describe how the `claude` CLI authenticates (it assumes user has installed/configured it).

✓ Install Mechanism

No install spec in registry (instruction-only plus bundled script). The SKILL.md recommends `npm i -g @anthropic-ai/claude-code` for the claude CLI, which is reasonable and expected. No remote downloads or extract operations are present in the included files.

ℹ Credentials

The skill declares no required env credentials (none listed), which is coherent because authentication happens via the locally installed `claude` CLI. However the script reads optional env vars (CLAUDE_RELAY_ROOT, CLAUDE_RELAY_MAP, XDG_* fallbacks) that are not documented in SKILL.md. Also, the skill will cause potentially sensitive conversation content to be written to local logs (~/.local/state/cc/logs) which may persist PII or secrets. Users should be aware the `claude` CLI itself typically needs an API key configured in the environment or config — the skill does not request or manage that key but will cause traffic to flow to that service.

ℹ Persistence & Privilege

always:false (normal). The script creates and writes configuration and state under the user's home directories (~/.config/cc and ~/.local/state/cc) and writes incremental logs. That behavior is expected for a relay but means persistent storage of forwarded content on disk; the skill does not modify other skills or system-wide settings.

版本历史

v2.1.0

cc v2.1.0 - The relay flow now uses only incremental output for user messages, not the full session history. - Large output message updates: the prompt for more detail now says "send /cc tail to see more". - ANSI escape codes are now stripped by the script automatically—no need to handle this in output formatting. - Documentation clarified to reflect the new script behavior and output requirements.

v2.0.1

Add ACP distinction note: cc continues existing claude -c sessions, unlike ACP which creates new ones.

v2.0.0

v2.0.0: Complete rewrite. Pure relay mode (removed discuss). Polling wait (up to 60s). /cc ? status check. ★ last-used project marker. macOS compatible. Safe config. POSIX-only (no mapfile). Auto language detection.

v1.1.0

Add relay mode transparent pipe contract with WRONG example, progressive disclosure (references/relay-mode.md), environment variables.

v1.0.0

- Initial release of cc skill: quick slash command wrapper for Claude relay sessions. - Provides /cc commands for starting/stopping sessions, tailing output, and messaging Claude Code via chat. - Supports project alias targeting, fuzzy project matching, and relay mode for auto-forwarding user messages. - Integrates with Telegram and other chat apps, including inline button support. - Requires claude-relay skill and dependencies (tmux, claude). - Emphasizes no narration or duplicate messages in chat output.

元数据

Slug cc

版本 2.1.0

许可证 MIT-0

累计安装 2

当前安装数 2

历史版本数 5

常见问题

Cc 是什么？

Claude Code relay via tmux. Operate Claude Code remotely — start sessions, send messages, read output. Use when the user wants to interact with Claude Code f... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 663 次。

如何安装 Cc？

在 OpenClaw 或 Claude Code 对话框中运行命令「/install cc」即可一键安装，无需额外配置。

Cc 是免费的吗？

是的，Cc 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。

Cc 支持哪些平台？

Cc 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。

谁开发了 Cc？

由 XING（@artwalker）开发并维护，当前版本 v2.1.0。

Cc