Troubleshooting

Start with the visible error and your current platform. Keep passwords, API keys, account-linking codes, pairing URLs, and private workspace content out of support reports. Use a redacted support bundle when one is available.

Workspace integrity or database permission warning

If diagnostics report an orphan workspace relation, preserve the workspace and contact your operator or Telemus support. Do not repeatedly run broad repairs or manually remove database rows to clear the warning. A repair can resolve one warning without resolving every recovery issue.

If startup reports unsafe database permissions, stop the affected application and contact the operator for the supported recovery procedure. Do not delete recovery files, follow an unfamiliar storage path, or change permissions across an entire folder. Keep backups and diagnostic snapshots private.

Use supported export and recovery tools for workspace data. A file copy of a running application's database is not a supported backup. Mobile and browser storage may require a different recovery procedure from a desktop installation.

Filesystem checkpoints are disabled for this space

Filesystem checkpoints are unavailable for a filesystem root, the host's user or service home, and folders that overlap the checkpoint store. The Web Checkpoints panel explains why Create is disabled.

1. Open Spaces → Add space.
2. Register an existing dedicated project folder that the host can access.
3. Select that project in Checkpoints, then review Create.

Do not disable the protection or move application storage to force a checkpoint. In the terminal, use /workspace-add <path> to register the folder and /rollback <path> to inspect checkpoints. Spatial room snapshots and filesystem checkpoints cover different information. Android, iOS, and browser-only clients need a supported desktop or headless companion for filesystem checkpoints.

Account sign-in and device recovery

An expired sign-in handoff returns to sign-in controls with a retry message. Sign in again on that page. If the device-link code has expired, return to the app or dashboard and start a fresh link.

When the Web dashboard requests account recovery:

1. Choose Link this device or Relink device.
2. Complete the portal steps.
3. Choose Check link status.

If no tab opens, use Continue account linking. If no link is available, retry rather than treating the device as paired. The account setup dialog covers background controls while it is open; choose Explore Command Center to return to them.

Relinking does not grant a licence or remove an update requirement. Update the app when requested, then retry. Dashboard access/password sign-in and Telemus account linking are separate.

If FINISH on onboarding asks you to reconnect your account, complete the Device Link steps and select FINISH again. Install a current build if the linking code does not appear or cannot be refreshed.

If a text field cannot accept input, report the page, field, and visible dialog. A disabled Send button can instead indicate an account, licence, or update requirement. Do not include authentication details in the report.

Companion relay says it is still starting

Channel connectors on desktop can depend on a local companion relay. A relay that is starting may become available after a short wait.

• A licence refusal requires an active licence; waiting does not resolve it.
• A local authentication failure requires channel setup to be completed again.
• A continuing startup failure should be reported with redacted diagnostics.

Spatial views and input

• Input does not respond: check the active panel, input mode, and controller or voice settings. Close an obstructing dialog before interacting with the view behind it.
• The saved environment is missing: install a current build, check that its asset is available, and use “list environments” followed by set environment to <name>. Say “show the grid” if you want the grid visible.
• A panel or text renders incorrectly: restart with a current build. Report the panel, platform, and whether the issue occurs every time. Missing or damaged fonts or avatars may require reinstalling the application.
• Quest slows down after a panel closes, or a panel's buttons stop responding: install a current build and retry. Report persistent failures with the panel name and a redacted support bundle.
• A rendering or media operation fails: use an offered external-viewer fallback where available. A supported platform does not guarantee that every file, driver, or remote service is compatible.

Exports, files, and recovery

• An export cannot be found: inspect the export result and destination shown by the application. Check that the destination is available and writable.
• An export refuses sensitive metadata: keep the underlying workspace private and contact support. Do not alter linked identifiers or share a database to bypass the refusal. Existing backups may still contain the older information.
• A cloud or network file fails to open: verify the configured provider, account access, connection, and supported file type. Browser-only clients do not gain direct access to every native file service.
• A room did not resume as expected: check the active workspace and the resume summary. A saved layout and an applied layout can have different states; follow the reported recovery guidance before making further changes.
• An undo or redo reports busy: let the current operation finish before retrying. Resetting history does not cancel changes already being applied.

Commands and approvals

A command containing template text such as your_ssh_target may be rejected. Supply the actual intended destination and review the new request normally; replacing template text does not grant permission to execute it.

Some commands need approval even when their description sounds read-only. Scripts, remote commands, and shell wrappers can perform additional actions. Review the proposed command and your active permission policy instead of disabling safeguards merely to remove a prompt.

Use the Web Approval policy card or terminal /policy to inspect effective settings. A setting supplied by the server may require an operator to change it.

A channel turn stops before finishing

When a task reaches a limit or cannot finish, Telemus may send an incomplete progress reply with available findings and a suggested next action. A successful individual tool call does not mean the entire request is complete.

Reply continue to request more work using the conversation context still available, or narrow the request. This is not a guarantee that the full research result or every earlier step can be recovered. History limits and conversation reset still apply. Separately saved artifacts remain available according to normal workspace retention.

Longer running time does not extend access permission or consent. Approval requirements continue to apply when work resumes.

Steering or stopping channel work

Send /steer <guidance> in the same chat as an active task. If the task has finished or cannot accept more guidance, Telemus reports that it was not queued. Steering takes effect between work steps and may not interrupt an external operation immediately.

Use /stop or /new in the relevant chat. These actions target that chat's work. Where the requester is verified, they can also cancel that requester's delegated helpers from the same conversation. They do not cancel another user's helpers or unrelated sessions. Independently started background processes may need to be stopped separately; running external operations can take time to return.

Support drafts and escalation

BUG DRAFT, FEEDBACK DRAFT, and CHECK DRAFTS in native Settings save text locally; they do not submit it. Use Activity EMAIL SUPPORT or the drawer Email Support Bundle for a diagnostics-email draft. Copy your saved text into the email, review it, and send it from your email client. Web and terminal support email actions use the same separate handoff.

Before contacting support:

1. Check the application's service, account, and licence status.
2. Confirm the active workspace, input mode, and relevant settings.
3. Record the visible error and what you were doing when it occurred.
4. Collect a redacted support bundle if available and review it before sharing.

Include the application version, platform, and reproduction steps. Never send passwords, tokens, pairing URLs, or an unreviewed workspace database.