Claude Forge
A ready-made Claude Code setup: agents, slash commands, skills, safety hooks, rules and MCP connections in one install
Install
/plugin marketplace add sangrokjung/claude-forge
/plugin install claude-forgeCommands and some skills only.
This is third-party code. Review the repository files before installing.
What it does
Claude Forge installs into Claude Code a set of specialist agents (planner, security reviewer, TDD guide, adversarial reviewer), slash commands such as /plan, /tdd and /code-review, skills and behavior rules. Background hooks block risky commands and secret leaks, resume after API errors, catch doom loops and write a session time report at the end. Since v4.0 every behavioral change is checked by a second agent that never saw the first one's reasoning. Skills include a session relay for handoffs, systematic debugging and trimming always-loaded context.
Who it is for. For developers who want a configured Claude Code with process and guardrails instead of assembling a config by hand.
Good fit when
- You want a connected plan, test, review and verify flow in Claude Code
- You need hooks that block dangerous commands
- You want changes independently checked by a second agent
Not a fit when
- You already have a tuned set of hooks and rules that would conflict
- You need a single skill without touching global settings
Example request
/plan add CSV export for reports, then /tddLimitations
The plugin install gives only commands and some skills; agents, hooks, rules and MCP come from install.sh, which changes the global Claude Code config. Requires Claude Code 2.1.110 or newer. Some skills target Korean text.
How to disable. Remove the plugin with /plugin uninstall claude-forge. After a full install remove the added agents, hooks and rules from ~/.claude and restore your previous settings.json.
Security check
- The installer changes global Claude Code settings and hooks
- Hooks run commands on every agent action
README in short
The README pitches the project as oh-my-zsh for Claude Code and explains the kit in plain terms. It covers two install options with a capability table, the agents, commands, skills and hooks, the architecture and an FAQ. Release notes describe adversarial verification, a reliability pack and session time reports. MIT licensed.
FAQ
How does the plugin differ from install.sh?
The plugin adds commands and some skills. The script adds agents, hooks, rules and MCP connections.
How do I update?
The plugin updates with /plugin update claude-forge; a full install with git pull and ./install.sh --upgrade.
Related
A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review
Skills for real engineers by Matt Pocock
Skills For Real Engineers
Small composable skills for engineering with agents: plan grilling, TDD, bug diagnosis, code review and architecture
Reference MCP servers
Model Context Protocol servers
Official reference MCP servers: Filesystem, Fetch, Git, Memory, Sequential Thinking, Time and Everything
Up-to-date, version-specific library docs and code examples in your agent's context, via MCP or a CLI plus skill