agentic-memory-vault

A structured, agent-first knowledge base template for organizations using AI agents alongside human teams.

What Is This

Memory Vault is a git-based knowledge management system designed for AI agents (Claude Code, Codex, Copilot, Cursor) and human teams. You clone this template, initialize it for your organization, and start building institutional memory that both humans and agents can read, write, and reason over.

The architecture is inspired by the Karpathy LLM Wiki pattern, extended for multi-agent enterprise use. Source documents go in, structured knowledge comes out. Agents process sources into wiki pages, answer questions from the knowledge base, and maintain vault health through automated linting.

No databases, no external services. The core workflow needs only bash, git, and standard Unix tools; python3 is required for the skill-hardening and content-policy tooling (without it, those checks warn or fall back to grep, and skill-audit / content-audit do not run). The vault is a directory of markdown files with YAML frontmatter, linked together with wikilinks, and governed by enforceable rules.

Quick Start

Full setup walkthrough: docs/getting-started.md

How It Works

Agents perform exactly three operations on the vault: