MemoryGuard docs

Practical notes for the public open-source alpha. MemoryGuard runs locally from source today and does not require a hosted service.

Install the memoryguard command

shell
# Windows (PowerShell)
irm https://raw.githubusercontent.com/atharvmantri/MemoryGuard/main/scripts/install.ps1 | iex

# macOS / Linux
curl -fsSL https://raw.githubusercontent.com/atharvmantri/MemoryGuard/main/scripts/install.sh | bash

memoryguard doctor
memoryguard demo

The full walk-through is on the Quickstart page. The one-liner clones the repo into a stable source directory, runs uv sync, writes a thin wrapper, and adds it to your PATH. After it finishes, daily commands are memoryguard init, memoryguard remember ..., memoryguard sync, and so on — no uv run in front of every call.

Quickstart

Install MemoryGuard in one command and use the memoryguard command directly.

Agent Capture

Turn Codex, Cursor, Claude, or plain session files into reviewable memory candidates.

Context Sync

Render approved project memory into AGENTS.md, CLAUDE.md, Copilot instructions, MEMORY.md, and Cursor rules.

Security

Local storage, deterministic redaction, and the alpha limitations you should know about.

CLI surface

The alpha CLI covers the daily local workflow:

shell
memoryguard doctor
memoryguard init
memoryguard remember "This project uses Flask for the backend."
memoryguard capture file ./session.txt --source codex
memoryguard capture pending
memoryguard capture approve --all
memoryguard sync
memoryguard status
memoryguard demo

What MemoryGuard does today

MemoryGuard keeps the context files used by AI coding tools aligned with approved project memory. The alpha includes:

  • A local SQLite memory store under .memoryguard/.
  • Context Sync for the files coding tools already read.
  • Obvious supersession handling (FastAPI → Flask, etc.).
  • Best-effort deterministic secret redaction.
  • A review-first Agent Capture workflow with pending approval and memoryguard capture reject.

How to think about MemoryGuard

MemoryGuard is not a general chatbot memory layer and not a full development-history system. It is a focused local tool for turning reviewed project decisions into maintained agent context files. It runs locally, has no required hosted model, and stores state in a SQLite file under your project.