Architectural Memory
for AI Coding
"We cannot predict the future. But we can refuse to forget the past."
LORE turns any codebase into structured, queryable knowledge that your AI assistant can actually reason about. 13 deep analyzers. Zero hallucinations about your architecture.
How LORE Works
Three simple stages transform your raw code into AI-ready architectural knowledge.
Scan Your Codebase
LORE recursively walks your project tree, parsing imports, exports, types, and configuration files to build a structural map of your architecture.
Run 13 Analyzers
Deep analysis engines fire in parallel: coupling matrices, dependency graphs, hotspot scoring, entropy measurement, type safety evaluation, and more.
Inform Your AI
Via MCP, Claude, Cursor, or Windsurf queries LORE on-demand. Your AI now reasons about real structural data — not guesses.
Deep Code Intelligence
Purpose-built analyzers that understand your architecture before your AI assistant ever opens a file.
Hidden Coupling
Reveals implicit dependencies between modules that imports alone don't show. Catches silent breakage before it happens.
Type Safety Score
Evaluates TypeScript type coverage and strictness across your entire codebase. Know where your type shield has holes.
Hotspot Analysis
Identifies change-prone modules that accumulate technical debt. Focus refactoring where it matters most.
Dependency Graph
Maps the complete module dependency tree. Visualize how every file connects to every other file in your project.
Circular Dependencies
Detects import cycles causing subtle bugs, memory leaks, and making code impossible to reason about.
Breaking Changes
Flags potential breaking changes before they reach production. Protect your API contracts and public interfaces.
AI Recommendations
Smart improvement suggestions powered by analysis results. Actionable recommendations tailored to your codebase.
Coupling Matrix
Quantifies inter-module coupling strength. Spot tightly-coupled clusters that resist independent evolution.
Entropy Scoring
Measures code complexity and identifies high-entropy zones. Know which parts need the most attention.
Import Impact
Shows downstream ripple effects of every import. Understand what changes when you modify shared modules.
Gap Analysis
Finds missing error handling, testing gaps, and inconsistent patterns. Stop leaks before they flood.
Tooling Config
Validates build tools, linters, and project configuration. Ensure your toolchain is properly tuned.
Middleware Chain
Tracks request processing paths through your middleware stack. Debug routing and data flow issues.
See It In Action
Run lore status on any project and get instant architectural intelligence.
Works With Your AI Tools
Connect via Model Context Protocol. Set up in seconds.
Claude Desktop
Full MCP integration
Windsurf
Codeium's AI IDE
Cursor
AI-first code editor
Any MCP Client
Protocol compatible
Runs Everywhere Node.js Runs
LORE is built on Node.js — meaning it works on macOS, all Linux distributions, and Windows. Zero native dependencies.
All Linux
Ubuntu, Debian, Fedora, Arch, Kali, Mint, CentOS, RHEL, Alpine
macOS
Intel and Apple Silicon (M1/M2/M3/M4) — native support
Windows
Windows 10/11 — PowerShell and CMD fully supported
CI/CD
GitHub Actions, GitLab CI, Jenkins — use in any pipeline
Up and Running in 60 Seconds
Install LORE
Grab it from npm or try it instantly with npx.
npm install -g lore-mcp
npx lore-mcp initAdd to Claude Desktop
Add LORE to your MCP client configuration file.
{
"mcpServers": {
"lore": {
"command": "npx",
"args": ["-y", "lore-mcp"]
}
}
}Ask Claude About Your Code
Claude now understands your architecture. Ask deep questions.
"What are the hidden coupling risks?"
"Which modules are the biggest hotspots?"
"Show me all circular dependencies."Battle-Tested on Real Projects
Not a toy. LORE has been tested on 16 major open-source TypeScript projects with a 100% pass rate and zero crashes.
+ 8 more projects tested — zero crashes, 100% pass rate across all 16 test subjects.