🐚 docker-shell

A lightweight, dependency-free bash script that gives you an emulated interactive shell inside a Docker container — using nothing but curl, dd, od, and awk. No Python, no jq, no editor required.

Built for environments where you only have raw shell access and need to exec commands inside a Docker container via the Docker Engine API.

🧠 How It Works

Instead of using the Docker CLI, this script talks directly to the Docker Engine REST API over HTTP (typically exposed at http://192.168.65.7:2375 in Docker Desktop / WSL2 environments).

It:

1. Creates and starts an Alpine Linux container with your Windows C:\ drive mounted
2. For each command you type, it creates an exec instance via the API
3. Parses Docker's multiplexed binary stream output using pure dd + od — no Python needed
4. Prints the result back to your terminal like a real shell

Docker Multiplexed Stream

Docker wraps all exec output in a binary framing format:

[1 byte: stream type] [3 bytes: padding] [4 bytes: payload size (big-endian)] [payload...]

This script decodes that frame-by-frame using only standard Unix tools.

⚙️ Requirements

• bash
• curl
• dd, od, awk, grep, cut — all standard Unix tools
• Docker Engine API accessible at http://192.168.65.7:2375 (Docker Desktop on WSL2)

No jq, no Python, no nano, no editor.

🚀 Installation

git clone https://github.com/yourusername/docker-shell.git
cd docker-shell
chmod +x docker_shell.sh
bash docker_shell.sh

💻 Usage

================================================
  Docker Emulated Shell  [target: alpine_escape]
  Commands: start | status | run <cmd> | exit
================================================

docker> start
docker> status
docker> run ls /mnt
docker> run ls /mnt/Users
docker> run cat /etc/os-release
docker> run id
docker> exit

Commands

🗂️ Mount Layout

The container mounts your Windows C:\ drive (via WSL2) at /mnt:

⚠️ Notes

• Each run is a separate exec instance — state does not persist between commands. Use run cd /some/dir && ls to chain commands in one call.
• The script targets the container by name (alpine_escape), not by ID, so it won't accidentally exec into the wrong container.
• Output is parsed directly from the binary file /tmp/docker_out.bin using dd frame-by-frame.

🔧 Configuration

Edit these variables at the top of the script to match your environment:

DOCKER="http://192.168.65.7:2375"   # Docker Engine API address
CONTAINER="alpine_escape"            # Container name

📄 License

AGPLv3