lore

Long-term memory for your projects and your AI

Every agent picks up where the last one stopped.

lore is one memory for all your work: a memory for each project, a shared memory for your workspace, and pages you can share between projects. Your coding agents, Claude and ChatGPT read and write it on every device and with your team, so the next session starts knowing what the last one learned.

Create an account See how it works

handoff from codex · laptop · 2 days ago
recall decisions/auth-model.md

How it works

Your agents already do the work. lore keeps what they learn.

Nothing to write by hand. lore listens while your agent works and turns sessions into memory the next one can read.

  1. Capture

    Lifecycle hooks record what your agent does as it works: the prompts, the tools it runs, the decisions it makes. Secrets are stripped before anything is stored.

  2. Compile

    lore turns those sessions into Markdown pages: decisions, conventions, how a part of the code works. Every page is versioned, and you can read and edit it in the browser.

  3. Pick up

    The next agent, on any device and for anyone on the project, reads those pages over MCP: in your editor, or in Claude and ChatGPT. Stop mid-task in Claude Code, open Codex, and carry on with a handoff.

Shared memory

Memories that connect, not one folder per laptop.

Most agent memory lives in a folder on one machine. lore keeps every project's memory on a server you share, and lets those memories reach each other.

Shared across projects

Workspace memory reaches every repository in the workspace: write a rule or a convention once and all of them know it. Share a single page from one project into another and it stays live.

Personal and team

Start with a personal workspace only you can see. Make team workspaces and invite people by username, email or link, as readers, writers or admins. Each device has its own key, revoked on its own.

Agents

Works with the agents and chat apps you already use.

  • Claude Code
  • Codex
  • Cursor
  • Gemini CLI
  • OpenCode
  • VS Code Copilot
  • Claude
  • ChatGPT

Agents with lifecycle hooks feed memory as they work; anything that speaks MCP can read and write it, including Claude and ChatGPT through the lore connector. Memory written by one is read by all the others.

Get started

Your agents remember from the next session on.

Every account starts with a personal workspace only you can see. When your team is ready, open a team workspace and share your projects' memory with them.

Create an account Sign in
  1. Create an accountPick a username. It appears next to everything you write.
  2. Connect this computerRun curl -fsSL https://app.lorememory.ai/web/install.sh | sh. It signs this device in through your browser and gives it its own key.
  3. Work as usualThe installer sets up Claude Code, Codex, Cursor and Gemini CLI. From their next session they read and write the memory. In Claude or ChatGPT, add the lore connector.