Credentials, nodes and schedules
Can anyone read a secret I stored in a credential?
No one can read it in the console. A secret is written once and never shown on screen again, not even to you: only the tasks the credential reaches receive it. When you open the credential, its secret fields show dots (●●●●●●●●●●●); left empty, they keep what they hold. To change a secret, type the new value into the same credential. Never paste a secret into a task's description, a project's commands or its agent instructions: anyone who can view those can read it.
See How secrets work and Replace or delete a secret.
Who can use a secret I add?
Its Owner decides. It is Personal by default; Org Owners and Admins can pick another under Advanced.
See Add a secret and How secrets work.
What if the service I need has no card in Add secret?
Use one of the general cards in the Custom category: Generic API key stores a key, and Generic login stores a sign-in, for any service without a card of its own. Other secret opens the plain form with every type in Type. To define your organization's own REST API or MCP server once, so members connect to it like any other service, an Org Owner or Admin adds a Custom connector or Custom MCP server.
See Add a secret and Add a custom API or MCP server.
How do I keep a leaving member's secrets and model accounts?
Give each one a new owner. An Org Owner or Org Admin can move any secret or model account to Organization to share it with the team, or to another member. A secret can also go to System, for Jaah's own use only. Stored values, sign-ins and settings move with it, so nothing is typed in again.
See Change a secret's owner and Change a model account's owner.
Can Jaah itself read my credentials' secrets?
Jaah encrypts every secret it stores. With NodeVault, the secrets of a credential marked NodeVault are sealed with your organization's own key, which only your nodes hold, so a leak of Jaah's database, backups or keys can't reveal them. Your node opens a secret only when a task on it fetches the secret. NodeVault is part of some plans, and an Org Owner or Admin sets it up once under Settings, then Features, then NodeVault.
See How NodeVault works and Set up NodeVault.
Does a node added later get the NodeVault key?
Yes, automatically. A node that joins your organization after NodeVault is set up gets the key from a node that already holds it, as a copy only the new node can unwrap, so its tasks can open NodeVault secrets without you doing anything. Because it happens on its own, check each new node: on Settings, then NodeVault, every node that holds the key is listed with its Fingerprint. Compare it with the fingerprint the machine itself reports, and look for the delivery in the audit log.
See How NodeVault works and Set up NodeVault.
What's the difference between a credential and a model account?
A credential connects your tasks to the services their work touches, such as Jira, Slack, GitHub or AWS. A model account is what a runner signs in as to reach the AI model itself. Credentials live under Directory, then Secrets; model accounts under Directory, then AI Accounts.
See How secrets work and Model accounts.
Does draining a node stop the tasks running on it?
No. Drain stops the node taking new tasks and lets the ones it's running finish; then the node disables itself. Disable stops new tasks at once, and running tasks carry on too. Only Delete ends what's running: it removes the node and every worktree on it, and can't be undone.
See Drain or disable a node and How nodes work.
Why does a node say Not mounted for a NodeLink?
The node couldn't reach the NodeLink's storage. A node connects to the storage's mount target, which has a private address only, so a public IP on the node doesn't help: the node needs a private route to it, such as peering, a VPN or a subnet router, and the storage must let its traffic in on TCP port 2049. The reason shows under Not mounted on the NodeLink's Nodes tab. When it says the node couldn't find the file system by name, the node is outside the file system's network: set Storage › Mount target IP › Change and route the node to that address.
See Set up the network for NodeLink and How NodeLink works.
Why can't I see Automations?
Audit, Automations, the Schedules tabs on projects and the Tasks page, and the Automations tabs on nodes are shown only to Owners and Admins of your organization. Playbooks are under Settings, Customize. Ask an admin to create the automation for you, or to grant you the role.
See How automations work and How team and access work.
Where is the list of all automations?
Under Audit, on its Automations tab: it lists every automation in your organization. To see only the ones that apply to one project, open the project's page and select its Schedules tab; for one machine, open the node's page and select its Automations tab. All of these are shown only to Owners and Admins of your organization.
See How automations work and Automations.
Will a task an automation filed stop to ask me questions?
That depends on the involvement picked in the automation's task. Unattended, the runner decides everything itself and asks only for what only you can supply, such as a missing credential. Guided, it also asks before key decisions and waits for your answer, like any task you started yourself. Interactive, it asks like Guided, then waits for you to pick Finish or Request changes when the work is done. A hidden task that asks shows on the board and notifies you.
See How automations work and Answer a task's question.
Why has my scheduled automation stopped filing tasks?
A schedule runs one run at a time. If a run goes past its time limit and never ends, it keeps its place and every run after it is skipped. Jaah marks the schedule Stuck and tells your Owners and Admins. An Org Owner or Admin opens Audit, then the Automations tab, opens the automation and selects Release, then Release in the dialog. When the next run starts, Stuck goes away and Run now works again. If the schedule is paused, release it before you select Resume.
See Release a stuck schedule and How automations work.
What happens to an automation when its owner leaves?
It stops filing tasks. An automation runs as the person who owns it, so once they leave your organization every run is refused. On Audit, Automations, it shows ⚠ Owner no longer in this org. An Org Owner or Admin opens its ⋯ menu, or the buttons on its page, and selects Take ownership. Its runs file tasks again from the next one; Run now runs it once straight away. If you don't want it any more, select Pause or Delete instead.
See Take over an automation whose owner left and How automations work.
Why didn't an automation file a task for an event it saw before?
An event automation remembers each item it filed a task for, for 90 days, so the same item arriving twice files one task, not two. To let an item file again, for example after you deleted its task, an Org Owner or Admin opens the automation's Advanced tab, finds the item's row under Dedup entries, selects Clear and confirms. Then send the event again. The task it filed before stays as it was.
See Let an automation act on an event again and How automations work.
Does deleting a NodeLink delete its files?
No. Deleting a NodeLink unmounts its folder from every node at once, but the files stay on the storage behind it, in your cloud and still billed. To remove them too, delete the storage as a second step: under Storage on the NodeLink tab, select ⋯ on its row, then Delete, and tick the box that also deletes it in your cloud. Storage you picked as Existing, and Azure file shares, are never deleted in your cloud. Copy off any file you want to keep first.
See Delete a NodeLink and its storage and How NodeLink works.
Can a Jira issue start a task?
Yes, once an Org Owner or Admin connects your Jira Cloud site under Settings, Integrations,
Issue trackers, with Add tracker, and maps each Jira project to a Jaah project. Then adding the
jaah label to an issue starts a task in the mapped project; with a Jaah account connection, assigning
the issue to Jaah or mentioning @Jaah does too. Jaah comments on the issue as the task goes and moves it
to Done when its pull request merges. Jira Server and Data Center aren't supported.
See Start tasks from Jira and How automations work.
Can a NodeLink use a file system we already have?
Yes, an AWS EFS file system in an account you have a cloud connection to. Open the NodeLink from Directory, Networks, NodeLink, select the Storage tab and pick the Connection. Select Existing AWS EFS, pick the File system, Mount target and Access point, then select Save and Provision, and wait for ready. Jaah never deletes that file system, but it replaces your own statements in its access policy. Its security group must admit NFS, TCP port 2049, from your nodes. Choose the storage before the NodeLink's access point exists.
See Put a NodeLink on a file system you already have and How NodeLink works.
Can someone who isn't an admin use our model account?
Yes, once you share the account with them. In the sidebar, select Directory, then AI Accounts, the account's name and its Access tab. Select Add, pick people or groups in Who, and select Add. Their tasks may then run on it; it reaches nothing else in your organization. An account has no roles: everyone on the list may use it. Sharing needs the Org Owner or Org Admin role.
See Share a model account with people and How model accounts work.
Should I connect a service or paste its key?
Connect it when its screen offers Connect with the service's name, such as Connect Figma. You sign in at the service and allow access: there is no token to create or copy, the connection renews itself, and the secret opens already tested. Under Advanced, Permissions lets you untick what the service should not allow. Choose Use an API key instead when you already hold a key. A service marked Beta has no one-click Connect yet, so its screen offers Use an API key first. Use your own OAuth app is for signing in through an app your organization registered at the service itself.
See Connect a service in one click and Secrets.
Can I have two secrets for the same service?
Yes, one per account or site, such as a production and a staging Jira. Select Add secret, find the
service in the gallery and select Add another on its card, then give the new secret a Name that says
which account it is, for example jira/staging. Tasks find a secret by its name, so name the one you mean
in the task's description with its runner path, or attach only that one to the project.
See Add a second secret for the same service and Tell a task which secret to use.
Why does a node say No beside Start on boot?
The machine doesn't run systemd, the service manager most Linux machines use. That is usual for a node installed inside a container. The node still runs tasks, but it doesn't start by itself after the machine restarts, so start it again yourself. Hover No to read the reason. The same machine can't hold GitHub runners, MicroVM sessions, NodeLinks or more nodes. To get them, install the node on a regular Linux machine or a cloud machine instead.
See What a node can't do and Add a node.
Why is Attach greyed out on a node's NodeLinks tab?
The reason shows beside it. NodeLink isn’t on your plan. means your plan leaves NodeLink out. Any other sentence is about the machine: shared folders need systemd, a machine allowed to mount them and the EFS mount helper. A node inside a container usually has none of these, so it can't mount a NodeLink. Install the node on a regular Linux machine or a cloud machine to use one. If Attach is missing altogether, you need the org admin role.