Loremem Cloud
The hosted Lore service: accounts, organizations, projects, tokens, and preconfigured models.
Loremem Cloud is the hosted Lore service at https://app.loremem.com. You sign in, keep memory in projects, create tokens, and connect agents. The memory model is the same as on a self-hosted server.
Sign in
Sign in at https://app.loremem.com/login with GitHub or Google. Accounts that an administrator creates can also sign in with an email address and password. Loremem Cloud doesn't offer public email registration.
Your first sign-in creates a personal organization and a default project.
Organizations and projects
An organization contains projects. A project is a separate memory space with its own memory graph, recall analytics, Dream runs, settings, tokens, and skills.
Use the organization and project switchers at the top of the console to change where you are. Create a separate project for each product or codebase whose memory should stay apart.
Tokens
Each token belongs to one project. To create one:
- Open Install and enter a Token note, such as
laptop. The note only records where the token is used. - Select Create token.
- Copy the token and the install command. Both are shown only once.
Manage tokens on the Tokens page (/apitokens). It lists each token's name, prefix, last four characters, status, and last-used time. Select Revoke to disable a token. Tokens look like lm_your_token.
Connect agents
The Install page shows an install command with your API address and the new token filled in:
npx @loremem/cli install --base-url https://api.loremem.com --api-token lm_your_tokenRun the command on each machine where you use an agent, then restart each agent runtime. The CLI connects the runtimes it finds on the machine. Pass --channels to choose them yourself.
Models
Each project selects two preconfigured profiles in Settings → Models: an embedding profile for recall, and a runtime LLM profile for Dream. You don't need to enter API keys.
Changing the embedding profile rebuilds the project's views and recall indexes, and it uses model quota. Changing only the runtime LLM profile saves right away without a rebuild.
What is the same and what is different
The memory model, recall, Dream, and backup settings work the same way on both. These are the differences:
| Area | Self-hosted server | Loremem Cloud |
|---|---|---|
| Memory model | Nodes, paths, disclosure, glossary, priority, and history | Same |
| Recall | Your embedding endpoint and recall settings | Same settings, using the project's embedding profile |
| Dream | Runs on your View LLM | Runs on the project's runtime LLM profile. Schedules follow the service clock. |
| Backups | Local files and WebDAV. Run, export, and restore from Settings | Same schedule, retention, and WebDAV settings. The Backup tab has no run, export, or restore buttons. |
| Skills | Not available | Available. See Skills. |
| Organizations and projects | One memory space per server | Organizations contain projects, and each project is separate |
| Models | You enter the embedding and View LLM endpoints and keys | Preconfigured profiles. No API keys to enter. |
API and MCP endpoints
- API base URL:
https://api.loremem.com - Generic MCP endpoint:
https://api.loremem.com/api/mcp?client_type=mcp, with the headerAuthorization: Bearer lm_your_token
Security notes
- Create a token for each machine, or for each place it is used, and give it a note that says where it is.
- Revoke any token you no longer use. Check Last used on the Tokens page to find idle tokens.
- Never paste a token into a memory, a skill, a prompt, or a shared chat.
- The CLI sends a token only over HTTPS, or over plain HTTP to a loopback address such as
localhost.
To move memory from a self-hosted server into a project, see Import a self-hosted backup.