DeusData/codebase-memory-mcp README.md: Complete Documentation Guide

Yes, the repository contains a comprehensive README.md file that serves as the primary reference for users and developers, covering installation, architecture, 14 MCP tools, and performance benchmarks.

The README.md in DeusData/codebase-memory-mcp functions as the central documentation hub for this code-intelligence engine. It describes the tool as "the fastest and most efficient code-intelligence engine for AI coding agents" and provides everything from one-line installation commands to detailed schema definitions for the underlying graph database.

Project Overview and Performance Claims

The README opens with a concise description emphasizing extreme indexing speed and zero-dependency architecture. According to the source documentation, the tool can index the entire Linux kernel in approximately 3 minutes with query latency under 1 millisecond (lines 31-38).

Key performance highlights include:

• Single static binary design with no external dependencies
• 158 languages supported through a hybrid parsing system
• Graph-based intelligence using Cypher-style queries for code relationships

Installation Methods

The documentation provides multiple installation pathways, from automated scripts to package managers.

One-Line Installation

For macOS and Linux systems, the README specifies this command (lines 42-53):

curl -fsSL https://raw.githubusercontent.com/DeusData/codebase-memory-mcp/main/install.sh | bash

To include the optional 3-D visualization UI:

curl -fsSL https://raw.githubusercontent.com/DeusData/codebase-memory-mcp/main/install.sh | bash -s -- --ui

Alternative Installations

The README also documents:

• Homebrew, Scoop, and AUR package manager support
• Pre-built binaries for manual installation
• Complete source build instructions for customization

MCP Tools Reference

A critical section of the README details the 14 JSON-RPC tools exposed by the Model Context Protocol (MCP) server (lines 116-137). These tools are implemented in src/mcp/mcp.c and include:

1. index_repository – Scans and parses a codebase into the knowledge graph
2. search_graph – Pattern-based search across nodes (functions, files, classes)
3. trace_path – Call-graph tracing between functions
4. query_graph – Cypher-style graph queries for complex relationships
5. get_file_context – Retrieves semantic context for specific file locations

Graph Data Model

The documentation defines the schema used in src/mcp/index_supervisor.c for the knowledge graph (lines 142-152).

Node Labels:

• Project, Package, File, Function, Class, Variable

Edge Types:

• CALLS – Function invocation relationships
• IMPORTS – Dependency relationships
• HTTP_CALLS – Inter-service API calls

Configuration and Security

Environment Variables

The README documents runtime configuration options (lines 170-196), including:

• Cache directory locations
• Memory budgeting parameters
• Diagnostic logging levels

Supply-Chain Security

A dedicated security section (lines 602-620) explains the project's SLSA Level 3 compliance, including:

• Cosign signatures for binary attestation
• VirusTotal scanning integration
• Provenance verification for all releases

CLI Usage Examples

The README provides practical examples for the command-line interface implemented in src/main.c.

Index a repository and list available projects:

codebase-memory-mcp cli index_repository '{"repo_path":"/path/to/your/project"}'
codebase-memory-mcp cli list_projects

Search for functions matching a pattern:

codebase-memory-mcp cli search_graph '{"project":"my-project","name_pattern":".*Handler.*","label":"Function"}'

Execute Cypher-like queries:

codebase-memory-mcp cli query_graph '{"project":"my-project","query":"MATCH (f:Function) RETURN f.name LIMIT 5"}'

Configure automatic indexing:

codebase-memory-mcp config set auto_index true

Launch the UI server (if installed with --ui):

codebase-memory-mcp --ui=true --port=9749

Hybrid LSP Architecture

The documentation explains the two-layer parsing system implemented in the indexer (lines 226-259):

1. Tree-sitter – Initial parsing for syntax trees across 158 languages
2. Hybrid LSP – Language-aware refinement for type resolution and cross-references

This architecture is orchestrated in src/mcp/index_supervisor.c and enables the background file watching functionality found in src/watcher/watcher.c.

Summary

• The README.md file provides complete documentation for installing, configuring, and using codebase-memory-mcp
• It documents 14 JSON-RPC tools for repository indexing and graph queries
• Performance benchmarks show Linux kernel indexing in 3 minutes with sub-millisecond query latency
• Security documentation includes SLSA Level 3 provenance and Cosign signature verification
• Code examples cover one-line installation, CLI usage, and Cypher-style graph queries

Frequently Asked Questions

What installation methods does the README.md document?

The README documents one-line bash installation for macOS/Linux, PowerShell steps for Windows, package manager installations (Homebrew, Scoop, AUR), and comprehensive source build instructions. It also includes a special --ui flag for installing the optional React/Three.js visualization interface configured in graph-ui/vite.config.ts.

How does the README describe the tool's performance characteristics?

According to the documentation, the tool achieves extreme indexing speeds through a zero-dependency static binary design. Specific benchmarks include indexing the entire Linux kernel in approximately 3 minutes and maintaining query latency under 1 millisecond, supported by a high-performance JSON parser in internal/cbm/vendored/yyjson/yyjson.c.

What security guarantees are documented in the README?

The README details SLSA Level 3 compliance with Cosign signatures for binary attestation, VirusTotal scanning integration, and complete supply-chain provenance. Additional security policies are expanded in the companion SECURITY.md file, while third-party licenses are cataloged in THIRD_PARTY.md.

What is the graph data model used by the tool?

The documentation defines a knowledge graph with nodes for Projects, Packages, Files, Functions, Classes, and Variables. Relationships are modeled through edges including CALLS, IMPORTS, and HTTP_CALLS, enabling Cypher-style queries for code intelligence and cross-service analysis.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}