Claude Safe — DevOps skill for Claude Code
Sandboxed Claude Code launcher — runs Claude in Docker with tmux session management and git worktree isolation.
How to install Claude Safe
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open fastbytes/claude-safe and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Claude Safe does
Sandboxed Claude Code launcher — runs Claude in Docker with tmux session management and git worktree isolation.
Alternatives in DevOps
- Render Deploy — Deploy applications to Render's cloud platform using Git-backed services 14.6k ★
- Ship Release — Systematic release workflow for RTK: build verification, version bump, changelog update, git tag, and push to 11.9k ★
- WP CLI And Ops — WordPress CLI and operations management 902 ★
README
claude-safe
Sandboxed Claude Code launcher. Runs Claude Code inside Docker containers with tmux session management and git worktree isolation.
Architecture
host shell
└─ claude-safe
└─ docker run -d (container: sleep infinity)
└─ tmux
├─ window 1: claude ← main session
├─ window 2: bash ← user splits
└─ window 3: claude ← user spawns another
tmux runs **inside** the container. Every pane and window is sandboxed. `Ctrl-b d` detaches back to the host — the container keeps running. `claude-safe` reattaches.
Why
Claude Code has unrestricted shell access. Running it on a bare host means it can `rm -rf`, read secrets, install packages globally, or modify system configs. claude-safe puts it in a container with explicit mount controls:
- Workspace: read-write (your project only)
- ~/.claude: granular ro/rw (settings ro, session state rw)
- ~/.kube, ~/.ssh, ~/.gitconfig: read-only
- ~/Code: read-only cross-project reference
- Everything else: not mounted
Requirements
- Docker (OrbStack recommended on macOS, Docker Engine on Linux)
- Git
- Bash 4+
- jq
tmux is included in the container image — not needed on the host.
Install
./claude-safe install
The installer detects your environment, prompts for each option with sensible defaults, writes `~/.config/claude-safe/config.sh`, installs the script, and builds the Docker image. Run `claude-safe install` again anytime to reconfigure.
Manual install# 1. Build the image
docker build -t claude-safe -f Dockerfile.claude-safe .
# 2. Install the script
cp claude-safe ~/.local/bin/claude-safe
chmod +x ~/.local/bin/claude-safe
# 3. Copy config
mkdir -p ~/.config/claude-safe
cp config.example.sh ~/.config/claude-safe/config.sh
# Edit to match your setup
# 4. (Optional) tmux config for remote access
cp tmux.conf ~/.config/claude-safe/tmux.conf
# 5. Ve
Related Skills
Ohmyharness
Run any coding agent in a Docker sandbox with your rules, skills and MCP servers already inside. Git worktree
Hort
Sandbox any CLI AI agent so it runs unrestricted without touching your machine. Written in Rust, daemonless, n
Porta
Sandboxed runtime for AI agents. WASM isolation + OS-level sandbox. Run Claude Code, Python, or any command wi
Claude In Box
Portable Claude Code dev environment in a Docker container — multi-session, hook-driven, web-managed, with tra
Run Claude Docker
Run claude code in somewhat safe and isolated yolo mode.
Praktor
Multi-agent Claude Code orchestrator with Telegram I/O, Docker isolation, swarm patterns, and Mission Control
Related Agents
Tsk - AI Agent Task Manager And Sandbox
by dtormoen - A Rust CLI tool that lets you delegate development tasks to AI agents running in sandboxed Docke
Git Agent
Background git operations agent - commits, PRs, branch management, release workflows. Runs on Sonnet to free m
Agent Of Empires
Claude Code, OpenCode, Mistral Vibe, Codex CLI, Gemini CLI Coding Agent Terminal Session manager via tmux and