Forge Plugin Adds Governance to Claude Code: 22 Agents, Quality Gates, and Zero Config
What It Does
Forge is an open-source governance plugin for Claude Code that adds automated judgment to your AI-assisted development workflow. It provides 22 specialized agents, quality gates, health scoring, and gap analysis—all running as background processes without configuration files or runtime dependencies.
The plugin addresses three critical gaps in current Claude Code usage:
- Automated checks that catch problems before they reach production
- Knowledge skills that encode your project conventions once
- Validation hooks that run before and after every Claude Code task
Setup
Installation takes 30 seconds with two commands:
# Add the Forge marketplace
claude plugin marketplace add nxtg-ai/forge-plugin
# Install the plugin
claude plugin install forge
That's it. No build step, no config files, no runtime dependencies. The plugin installs as agent definitions, command templates, skill documents, and hook scripts that Claude Code reads directly.
Key Commands
Once installed, you get these new commands:
/forge:status→ Project health score at a glance/forge:gap-analysis→ Find missing tests, docs, and security gaps/forge:feature "desc"→ Multi-agent feature development/forge:checkpoint→ Save restorable project state/forge:test→ Run tests with detailed analysis/forge:deploy→ Deploy with pre-flight validation
How Governance Works
Forge adds governance hooks that run automatically:
- Pre-task hooks validate that work aligns with project constraints before Claude starts
- Post-task hooks check output quality after Claude completes tasks
This makes governance a background process rather than a manual checklist item. The hooks trace to real failure modes the creator observed—audit findings from shortcuts, repeated context explanations, and deployment-discovered bugs.
MCP Integration
The plugin includes 8 MCP (Model Context Protocol) governance tools that expose health scoring, code metrics, and security scanning. Any tool in your workflow that speaks MCP can query Forge's governance state, creating interoperability across your development toolchain.
When To Use It
Use Forge when:
- You're working on production code and need automated quality gates
- Multiple developers share Claude Code sessions and need consistent conventions
- You want to catch security gaps or missing documentation early
- You need to save project states that can be restored later
- You're deploying code and want pre-flight validation
Advanced Options
For multi-tool orchestration across Claude Code, Codex CLI, and Gemini CLI (with file locking, shared knowledge, and task boards), add the Forge Orchestrator:
curl -fsSL https://forge.nxtg.ai/install.sh | sh
forge init
This is a 4MB Rust binary with zero runtime dependencies that makes separate AI tools work as a coordinated team.
For visual dashboards and the Infinity Terminal (sessions that survive browser close, network drops, and server restarts), add Forge UI:
git clone https://github.com/nxtg-ai/forge-ui && npm install && npm run dev
Each layer builds on the last, and nothing forces you to go deeper than you need. Adoption follows the pain points you're experiencing.
Why It Matters
Claude Code agents already coordinate well internally, but they lack judgment. Forge adds that missing layer—automated checks that prevent problems from reaching production, skills that encode your conventions once, and hooks that validate work continuously. It's governance that works in the background while you focus on building.
