Install Loom, connect your agents, manage stores, and run Kanban. Step-by-step walkthroughs are in the tutorials.
Connect a coding harness to durmem, save one harmless project memory, and recall it in a fresh session. This route uses hosted durmem by default.
Start with the [quickstart](quickstart.md): install the local command, connect a coding harness, and save and recall one harmless project memory. This page explains the ideas behind that route and points to the next guide or practice exercise.
Use the [quickstart](quickstart.md) for a first save-and-recall. This page explains the installer, setup versus login, harness checks, and local update or removal.
Connect Hermes to durmem with an MCP server entry in `~/.hermes/config.yaml`. Hermes has no durmem plugin manifest, and `durmem setup` does not configure Hermes. This is an MCP-only setup: the bundle's config snippet does not register or enable lifecycle hooks, and this guide does not configure Hermes hooks. Follow the shared `loom-coding` discipline manually.
Create a durmem account, understand its limits, and request help or deletion of your data.
`~/.loom/stores.toml` tells durmem which stores you can reach, which to use by default, and which settings apply to a workspace.
Create, inspect, recover, or remove stores in the web app or CLI.
durmem Kanban is a shared task ledger for trusted agents. A board is a durmem store with the Kanban protocol initialized; tasks record bounded work, runs record attempts, and an owner is responsible for questions and task closure. If you work with one agent on one computer, you can skip Kanban.
A Kanban board is a dedicated durmem store whose domain is `LogDomain` and whose Kanban protocol has been initialized. Hosted and private sign-up create the `kanban` starter store; check for that store before making another. Keep it separate from your memory store so coordination records do not become project or personal memory.
A fixed-session worker is one harness-native session that an operating-system schedule checks for Kanban work. Set it up on the computer that will run it, from an authenticated interactive session in the intended repository. Installing a worker changes that computer's schedule and gives the unattended session broad access to its OS account and checkout; use only computers and repositories you trust. Claude runs with permission prompts bypassed; Codex runs with full filesystem access and no approval prompts. This is not a sandbox for untrusted tasks.
Fleet applies an operator's desired worker settings to trusted computers. A non-LLM residual on each computer reconciles local checkouts, harness sessions, and schedules. It is a control-plane reconciler, not a remote shell. Keep a residual on every computer that Fleet must be able to manage, including when its desired worker count is zero.
Find the durmem setup, maintenance, store-management and coordination commands. The `durmem` command (the CLI) connects this computer and its coding harnesses to your durmem server; it is not the server. For everyday memory, ask your connected harness or agent to save or recall facts; CLI memory verbs are for internal use.
Use the symptom below to choose a safe next step. Replace `durmem.ai` with your durmem server's address (for durmem, `durmem.ai`). If you share an error, redact sign-in codes, links, store keys, and the contents of `stores.toml`; issue reports may be visible beyond your team. For a first connection, see [Install and set up durmem](install.md).
Protect your credentials, understand who can access your data, and report security issues privately.