← Back to Skills Marketplace
smallnest

Acp Remote

by smallnest · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
95
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install acp-remote
Description
Connect to remote ACP server and execute commands via imclaw-cli.
README (SKILL.md)

ACP Remote

Connect to a remote ACP (Agent Client Protocol) server and execute commands using imclaw-cli.

Prerequisites

Auto-install dependencies

Before using this skill, ensure both imclaw-cli and acpx are installed. Run these commands to install them automatically:

# Install acpx (ACP protocol support)
if ! command -v acpx &> /dev/null; then
  npm install -g acpx@latest
fi

# Install imclaw-cli
if ! command -v imclaw-cli &> /dev/null; then
  curl -fsSL https://raw.githubusercontent.com/smallnest/imclaw/main/scripts/install.sh | bash
fi

Manual installation

Install acpx

npm install -g acpx@latest

Install imclaw-cli

From GitHub Releases:

# 1. Detect your platform
OS=$(uname -s | tr '[:upper:]' '[:lower:]')
ARCH=$(uname -m)
[ "$ARCH" = "x86_64" ] && ARCH="amd64"
[ "$ARCH" = "aarch64" ] && ARCH="arm64"

# 2. Get latest release version
LATEST=$(curl -s https://api.github.com/repos/smallnest/imclaw/releases/latest | grep '"tag_name"' | sed -E 's/.*"([^"]+)".*/\1/')

# 3. Download and extract
curl -sL "https://github.com/smallnest/imclaw/releases/download/${LATEST}/imclaw_${OS}_${ARCH}.tar.gz" | tar xz -C /tmp

# 4. Install to PATH
mkdir -p ~/bin
mv /tmp/imclaw-cli ~/bin/
chmod +x ~/bin/imclaw-cli

# 5. Ensure ~/bin is in PATH
echo 'export PATH="$HOME/bin:$PATH"' >> ~/.bashrc  # or ~/.zshrc

Verify installation

imclaw-cli --help
acpx --version

Configuration

Set the remote server URL and optional auth token:

# Default server (can be omitted if using default)
export IMCLAW_SERVER="ws://your-server:8080/ws"

# Auth token (if server requires authentication)
export IMCLAW_TOKEN="your-secret-token"

Or pass directly via command line.

Server Configuration

IMClaw server uses command line flags (no config file needed):

# Start server with default settings
imclaw

# Custom port and auth token
imclaw --port 9000 --token my-secret-token

# Show help
imclaw --help
Flag Default Description
-H, --host 0.0.0.0 Server host address
-p, --port 8080 Server port
--timeout 30 Default timeout in seconds
--token "" Authentication token (empty = no auth)

Usage

One-shot mode (recommended)

# Basic usage
imclaw-cli --server ws://your-server:8080/ws -p "Hello"

# With auth token
imclaw-cli --server ws://your-server:8080/ws --token your-token -p "Hello"

# Use specific agent
imclaw-cli --server ws://your-server:8080/ws --agent codex -p "Analyze this code"

# JSON output
imclaw-cli --server ws://your-server:8080/ws --format json -p "List files"

Continue a session

# First message creates session, returns session ID
imclaw-cli --server ws://your-server:8080/ws -p "Read the README.md"

# Continue with same session
imclaw-cli --server ws://your-server:8080/ws --session \x3Csession-id> -p "Summarize it"

Parameters

Parameter Description
--server WebSocket URL of ACP server (default: ws://localhost:8080/ws)
--token Authentication token (if required)
-p, --prompt Prompt message (one-shot mode)
--session Session ID to continue
--agent Agent type: claude, codex, etc. (default: claude)
--format Output format: text, json, quiet (default: text)
--approve-all Auto-approve all permission requests
--approve-reads Auto-approve reads, prompt for writes (default)
--deny-all Deny all permission requests
--allowed-tools Comma-separated tool names (default: Bash,Read,Write)
--cwd Working directory
--timeout Max wait time in seconds
--model Agent model ID

Examples

Execute shell commands on remote

imclaw-cli --server ws://remote-server:8080/ws \
  --agent claude \
  --approve-all \
  -p "Run 'df -h' and show disk usage"

Analyze remote codebase

imclaw-cli --server ws://remote-server:8080/ws \
  --cwd /path/to/project \
  -p "Review the main.go file and suggest improvements"

Use with SSH tunnel

# Create SSH tunnel first
ssh -L 8080:localhost:8080 user@remote-server -N &

# Connect via localhost
imclaw-cli --server ws://localhost:8080/ws -p "Hello from remote"

Secure connection with token

# Server configured with auth_token
imclaw-cli \
  --server wss://secure-server:8080/ws \
  --token "your-secret-token" \
  -p "Execute analysis task"

Environment Variables

For convenience, you can set these environment variables:

# In ~/.bashrc or ~/.zshrc
export IMCLAW_SERVER="ws://your-server:8080/ws"
export IMCLAW_TOKEN="your-token"

Then simply run:

imclaw-cli --server $IMCLAW_SERVER --token $IMCLAW_TOKEN -p "Hello"

Tips

  1. Session reuse: Save the session ID from first response to continue conversations
  2. Permission control: Use --approve-all for non-interactive automation
  3. Output parsing: Use --format json for programmatic consumption
  4. Tool restrictions: Use --allowed-tools to limit what the agent can do
  5. Working directory: Always specify --cwd for file operations
Usage Guidance
This skill appears to do what it says, but take precautions before use: 1) Avoid running the suggested auto-install command that pipes a script from raw.githubusercontent.com into bash—inspect the script first or use the manual GitHub release installation. 2) Do not use --approve-all unless you fully trust the remote ACP server and its agents; prefer finer-grained approvals. 3) Be careful with --cwd and storing IMCLAW_TOKEN in shared shells—these enable remote agents to read/write local files and use your token. 4) If you must try it, run it in an isolated environment (container or VM), inspect downloaded binaries/scripts, and limit network access and credentials accessible to the environment.
Capability Analysis
Type: OpenClaw Skill Name: acp-remote Version: 1.0.0 The skill facilitates remote command execution using the imclaw-cli tool and acpx protocol. It promotes high-risk security practices, including a 'curl | bash' installation pattern from a remote GitHub repository (smallnest/imclaw) and manual modification of shell configuration files (~/.bashrc). Additionally, the documentation encourages the use of the '--approve-all' flag, which bypasses interactive permission prompts for remote operations, significantly increasing the risk of unauthorized command execution.
Capability Assessment
Purpose & Capability
Name and description (connect to remote ACP server and execute commands via imclaw-cli) align with everything in SKILL.md: required binaries are imclaw-cli and acpx, usage examples and flags match the stated purpose.
Instruction Scope
Instructions stay within the feature boundary (install tools, configure server URL/token, run imclaw-cli). However the docs explicitly recommend using --cwd (allowing remote operations on arbitrary local directories) and --approve-all (auto-approving permission requests). Those are expected for a remote-execution client but significantly broaden what a remote agent can do if used carelessly.
Install Mechanism
The SKILL.md advises auto-installation that pipes a script from raw.githubusercontent.com into bash (curl ... | bash). Even though the host is GitHub, piping remote scripts to a shell is a high-risk practice. Manual install via GitHub releases is safer but the provided auto-install pattern is still present and could be executed by less-knowledgeable users or agents.
Credentials
No unexpected credentials or unrelated environment variables are required. The document references IMCLAW_SERVER and IMCLAW_TOKEN for legitimate configuration; requesting an auth token for a remote server is proportional to the skill's purpose.
Persistence & Privilege
The skill is instruction-only, always:false, and not requesting persistent platform privileges. Still, combining agent autonomous invocation (platform default) with use of --approve-all and letting the remote agent operate in an arbitrary --cwd increases operational risk; this is a behavioral/operational concern rather than an incoherence in the skill itself.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install acp-remote
  3. After installation, invoke the skill by name or use /acp-remote
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release of acp-remote - Connects to remote ACP servers and executes commands using imclaw-cli. - Auto-installs and verifies required dependencies: imclaw-cli and acpx. - Provides setup instructions for both automatic and manual installation. - Details configuration options, usage parameters, and sample command-line usage. - Supports authentication, SSH tunneling, file operations, and session management. - Includes tips for session reuse, permission control, and output formatting.
Metadata
Slug acp-remote
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Acp Remote?

Connect to remote ACP server and execute commands via imclaw-cli. It is an AI Agent Skill for Claude Code / OpenClaw, with 95 downloads so far.

How do I install Acp Remote?

Run "/install acp-remote" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Acp Remote free?

Yes, Acp Remote is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Acp Remote support?

Acp Remote is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Acp Remote?

It is built and maintained by smallnest (@smallnest); the current version is v1.0.0.

💬 Comments