Iterare LLM — DevOps skill for Claude Code
Automated, isolated execution of Claude Code prompts in firewall-restricted Docker containers.
How to install Iterare LLM
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open sohonetlabs/iterare-llm and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Iterare LLM does
Automated, isolated execution of Claude Code prompts in firewall-restricted Docker containers.
Alternatives in DevOps
- Coi — Give the agent a machine 743 ★
- Cloudflare Skill — Comprehensive Cloudflare platform reference docs for AI/LLM consumption 709 ★
- Code On Incus — Give each AI agent its own isolated machine with root, Docker, and systemd 670 ★
README
Iterare
Automated Claude Code execution in isolated Docker containers.
`iterare` creates git worktrees, launches containerised Claude Code instances with autonomous permissions, and maintains safety through network firewalls and container isolation. This enables "agentic loops" where Claude Code can modify codebases without human confirmation while being restricted to whitelisted network access and mounted files.
_**Warning – This does not fully protect from malicious code execution or data exfiltration. It simply raises the barriers to these things happening. Use at your own risk.**_
Prerequisites
- Python 3.13+
- uv for dependency management
- Docker
- A Claude Code API key
Installation
# Install via pypi
pip install iterare-llm
# Create global config directories
iterare install
# Fetch Claude Code credentials (interactive Docker session)
iterare credentials
Quick Start
The easiest way to get started is to use the interactive mode. This launches a Claude Code session in a Docker container, mounts your current project directory, and allows you to drive Claude Code yourself.
# Initialize a project
cd /path/to/your/project
iterare init
# Launch in interactive mode
iterare interactive
Alternatively, you can create a prompt file and execute it. This (by default) creates a git worktree, and branches off of the current branch. It then launches Claude Code in a Docker container, mounts the worktree directory, and passes the prompt to Claude Code which will then execute the prompt until it's done. This is fantastic for autonomous development.
# Initialize a project
cd /path/to/your/project
iterare init
# Create a prompt
cat > .iterare/prompts/refactor.md << 'EOF'
---
workspace: refactor-auth
branch: main
---
Refactor the authentication module to use JWT tokens.
Add tests for the new implementation.
EOF
# Execute the prompt
iterare execute refactor
# Monitor progress
iterare log -f
Related Skills
Automated Flywheel Setup Checker
Automated testing framework for ACFS installer scripts — runs all 41 installers in isolated Docker containers
Runwall
The active firewall and execution governance gateway for AI agents. Run Cursor, Claude Code, Cline, and custom
Agentor
Self-hosted alternative to Claude Code Web, Codex in the Cloud, and similar managed agent environments. Spawns
Nanoclaw Aio
Telegram-first All-in-One Unraid container for NanoClaw. Runs Claude-powered agents in isolated Docker contain
AI Dev Container
devc creates isolated, sandboxed Docker containers for AI coding agents (Claude Code, Codex, Gemini CLI, Openc
Apex Specagent
Sovereign Speculative Tool Execution & Persistent Radix KV-Cache Engine for NVIDIA TensorRT-LLM, NIM & Blackwe
Related Agents
Docker Sandbox
Isolated autonomous scanning agent — runs in a temporary worktree with restricted Bash access. Requires ROE au
Amoa Docker Container Expert
Manages Docker containers for cross-platform development and isolated testing. Requires AI Maestro installed.
DevOps AWS Expert
Use this agent when the user needs help with DevOps tasks involving AWS services, Docker containers, Terraform