Welcome to Massu AI

Massu AI is an AI Engineering Governance Platform that transforms how your team works with AI coding assistants. It is an MCP server and Claude Code plugin that provides cross-session memory, a structured knowledge system, automated quality tracking, compliance audit trails, and 72 specialized tools -- all running locally with zero cloud dependencies.

What is Massu AI?

When you use AI coding assistants like Claude Code, every session starts from scratch. The AI has no memory of what happened yesterday, no knowledge of which approaches already failed, and no awareness of your team's coding standards. Massu AI fixes this.

Massu AI sits between your AI assistant and your codebase, providing:

• Cross-session memory that persists decisions, failures, and context across sessions
• Knowledge system with full-text search across rules, incidents, patterns, and verifications
• 11 lifecycle hooks that automatically capture observations, inject context, and enforce rules
• 72 MCP tools for code intelligence, security scoring, cost tracking, knowledge search, and more
• 15 canonical rules that ensure verified, auditable AI-generated code
• Config-driven architecture that adapts to any TypeScript or JavaScript project

Who is Massu AI For?

Massu AI is built for developers and engineering teams who use AI coding assistants at scale. If any of these describe your situation, Massu AI is for you:

• You use Claude Code (or plan to) and want your AI assistant to remember context across sessions
• Your team needs governance and compliance tracking for AI-generated code changes
• You want automated quality scoring that tracks improvement (or regression) over time
• You need to know what your AI assistant costs per feature, per session, and per model
• You care about security and want automated scanning of AI-generated code
• Your team collaborates on the same codebase and needs conflict detection

What Makes Massu AI Unique

Memory That Persists

Most AI coding tools treat every session as a blank slate. Massu AI maintains a local SQLite database that stores observations, decisions, failed attempts, and session summaries. When you start a new session, the session-start hook automatically injects relevant context -- including warnings about approaches that already failed.

Zero-Effort Automation

Massu AI's 11 lifecycle hooks run automatically. You do not need to remember to log decisions, capture observations, or save session state. The hooks handle all of this in the background, completing in under 500ms to avoid slowing down your workflow.

72 Purpose-Built Tools

Massu AI provides 72 MCP tools organized across eighteen categories. Free-tier tools (12) are available to everyone; additional tools unlock with Pro, Team, and Enterprise tiers:

Config-Driven Architecture

Everything is configured through a single massu.config.yaml file. Define your project's framework, domains, coding rules, security thresholds, quality weights, and team settings in one place. Massu AI adapts its behavior to your project.

Memory System Setup

When you run npx massu init, Massu AI automatically configures the memory system, installs all 11 lifecycle hooks, and deploys 31 slash commands into .claude/commands/. No additional setup is required. The session-start hook begins injecting context from your very first session. To check memory is working:

npx massu doctor
# Should show: Memory DB: OK

Optionally, create a MEMORY.md file in your project root to provide persistent background context that is injected at every session start. See Memory & Learning for details on the MEMORY.md format and topic files.

Knowledge System Overview

The knowledge system provides a structured, full-text-searchable store for everything your team learns -- rules, incidents, patterns, verifications, corrections, plans, and effectiveness measurements. Unlike the memory system (which records what happened in sessions), the knowledge system records what your team knows.

The knowledge system is backed by an SQLite FTS5 index and is accessible through 12 MCP tools:

• Search: massu_knowledge_search -- find anything across all knowledge types
• Rules: massu_knowledge_rule -- store and retrieve coding rules
• Incidents: massu_knowledge_incident -- record and query past incidents
• Patterns: massu_knowledge_pattern -- capture reusable development patterns
• Verification: massu_knowledge_verification -- track verification results
• Schema checks: massu_knowledge_schema_check -- record schema validation findings
• Graph: massu_knowledge_graph -- explore cross-references between knowledge entries
• Commands: massu_knowledge_command -- surface relevant commands for a context
• Corrections: massu_knowledge_correct -- record and retrieve AI corrections
• Plans: massu_knowledge_plan -- store and search plan documents
• Gaps: massu_knowledge_gaps -- identify knowledge coverage gaps
• Effectiveness: massu_knowledge_effectiveness -- measure knowledge utilization

See Knowledge System for full documentation.

Verification-First Workflow

Massu AI includes 31 workflow commands that enforce a verification-first approach to AI-assisted development. Every claim of completion requires proof. Every removal requires negative verification. Every plan item gets tracked.

Quick Navigation

How It Works

Your Prompt --> Claude Code --> Massu AI Hooks --> Memory DB + Knowledge DB
                    |                                          |
                    v                                          v
              72 MCP Tools <-- Context Injection <-- Past Sessions + Knowledge
                    |
                    v
              Governed, Verified, Auditable Code

1. You write a prompt in Claude Code
2. The user-prompt hook captures it for search and task linking
3. Claude Code calls tools -- Massu AI's 72 MCP tools provide code intelligence, memory search, knowledge lookup, quality scoring, and more
4. The post-tool-use hook captures observations from every tool call
5. The post-edit-context hook surfaces relevant rules when files are edited
6. When the session ends, the session-end hook generates a summary and archives state
7. Next session, the session-start hook injects all that context back in

The result: your AI assistant gets smarter with every session, your code quality improves over time, and you have a complete audit trail of every AI-assisted change.

System Requirements

• Node.js 18 or later
• Claude Code CLI (installed and authenticated)
• Git repository (Massu AI uses git for branch detection)
• Operating System: macOS, Linux, or Windows (WSL)

Ready to get started? Head to the Installation Guide.