Troubleshooting

Check sign-in, invitations, AI connections and storage, then investigate backup and restoration without losing records. Read any displayed error first and follow the relevant section.

Sign-in and accounts

1. For website membership, check the current account on the membership page; for archives and AI connections, check Account in the app.
2. Choose a social sign-in option actually offered on the screen. Confirm the browser’s social account is the one you want.
3. If verification expired or a network error occurred, authenticate again with the same account. An unverified offline account display is not completed server authentication.
4. During registration, confirm acceptance of the terms. Marketing notifications are optional.

Different Alfred accounts do not merge automatically even with matching email addresses. If materials disappear after switching accounts, check the original account and archive first. Confirm the current storage before creating a new archive or importing again under another account.

If withdrawal is denied, check for archives/organizations you own. Resolve the ownership-transfer or deletion requirements shown. Pressing withdrawal alone is not completed deletion. See membership guidance.

Invitations and permissions

An organization owner invites teammates by email on the organization screen. No invitation email is sent, so tell them directly. They can accept using an account with that email verified.

1. Confirm the invited email matches your Alfred account’s verified email.
2. Check and accept received invitations under Organizations in the app. Issuing and accepting an invitation are separate steps.
3. Choose an authorized organization archive after acceptance. Creating or joining an organization does not create a new archive automatically.
4. If you can read but not edit, ask the owner about your current organization/archive role. Viewers are read-only.

Existing AI connections cannot access archives after permission is revoked. Check archive reading permission, AI access settings and OAuth scopes separately. Being an organization owner does not grant access to someone else’s personal archives or private-feedback operator rights.

Records and storage

1. Check the selected archive name and storage location.
2. For browser-local materials, use the same browser, profile and site where you wrote them. For PC-local materials, use that device’s Windows app.
3. For server materials, complete sign-in verification with the right account and select the correct personal/organization archive.
4. Check diary/wiki, search, period and classification filters. Use Expand document list if collapsed.
5. If a save error appeared, reopen to check whether saving actually occurred before retrying.

Can I use it without signing in?

Yes. Device/browser-local archives work without sign-in. Server archives, organization sharing and server MCP need sign-in. Local archive data remains independent and is not automatically uploaded.

Browser-local records can disappear after deleting site data or changing browser profiles. Do not assume sign-in restores them from a server. Do not clear browser data first to solve a problem.

If documents exist but dashboard counts are empty, check real classification, date and schedule connections. Document totals and group memberships are not the same count.

AI connections

1. Confirm the server archive address is https://alfred3x5.com/mcp. Use current settings displayed by the app for PC-local MCP.
2. For server connections, check the current account, archive AI permission and granted OAuth scopes.
3. Refresh server status and permissions in External AI connections. Check that the external plan and organization allow custom MCP.
4. For PC localhost HTTP, ensure the Windows app is open and HTTP is running. After restarting, recopy configuration including the changed token, even if the port is unchanged.
5. If reading works but editing fails, check current account edit permission, archive write permission and the connection’s write grant. Read-only approval does not expand automatically.

If automatic copying fails, select and copy the address manually. See the address and service-specific steps in the AI connection guide.

Entering localhost in a ChatGPT/Claude web connector differs from local MCP on the same PC. Browser-local archives are not shared automatically with Windows-local archives.

Backups and importing

Keep a backup first

1. Select the archive and confirm its name and storage.
2. Select App menu › Settings › Export backup.
3. Confirm the downloaded JSON belongs to the expected archive and keep it elsewhere. Existing files and retained history are included in full backups.

Backups can contain sensitive source materials. Do not put them in public links/shared folders or attach them wholesale to inquiries. Server encryption at rest does not guarantee encryption of all downloaded local backups.

Restore a backup

1. Preserve current materials first and check the backup’s archive ID, storage and format.
2. Select its JSON in App menu › Settings › Import backup.
3. Confirm the destination shown by the app. Restoration is only to an empty store with the same archive ID, not an arbitrary merge or overwrite.
4. After success, check records, retained history and required links.

Server backups can be restored only to the original server archive. Encrypted server backups cannot be imported into local archives. Current permission to the destination is also required.

If a local backup’s archive ID differs from the open archive, the app can open the matching local archive and attempt restoration. This does not clone materials to another ID or merge two archives.

Do not delete existing records or browser data first to restore. If the destination is not empty or the original archive is inaccessible, stop importing and check the target and permission. Downloading a backup does not prove it can be restored.

Revision history and restoration

Document revisions are retained, so with edit permission you can restore an earlier version. Deleting or emptying the archive itself cannot be undone.

1. Open the text record and inspect retained versions in Revision history.
2. Open the earlier content and confirm it is the right revision. A deletion revision cannot be a restoration source.
3. With edit permission, choose and confirm Restore this version.
4. Reopen the new current version and check its contents.

Content not retained in history cannot be invented and restored. Distinguish archive deletion/emptying from restoring one text record’s revision. Do not apply this text restoration procedure to suspended file features.

Frequently asked questions

How do I organize materials?

Connect diary and wiki records through period, relationships, projects, resources and topics. Follow titles and body links to related content and narrow lists with filters. Group counts require explicit membership links.

What happens to existing attachments?

Existing attachments are retained safely without deletion and included in full backups. Adding, opening and previewing new files is temporarily suspended. See the current file feature scope.

How are dated records created?

Write events and plans/schedules directly in the diary, or ask AI (MCP) to review and register explicit document schedules under the scheduling rules. Passing a scheduled date alone does not automatically create an actual event.

Where do documents go when I connect ChatGPT or Claude?

Content AI searches or reads in authorized archives is sent to the connected ChatGPT/Claude service. This can include tool responses such as titles, sources and search results, not only document bodies.

Disconnect any time under [App menu › Account › External AI connections]. The server immediately blocks that connection’s access. (Content already transferred to the external AI service is not recalled.)

If the issue remains

Sign in with your existing member account in Feedback to report bugs, suggest features or ask questions. Only authors and designated operators see inquiry titles, bodies, status and replies.

• Include your environment, the problem, steps taken and displayed error.
• Add explanations as follow-ups to your inquiry. Existing messages cannot be edited or individually deleted.
• Feedback does not grant access to your whole archive or automatically send original materials to AI.
• Private means an access policy, not end-to-end encryption that even the server cannot read.
• Your own inquiries and conversations are deleted together when the existing account deletion procedure completes. Conditions such as archive/organization ownership still apply. See the privacy policy for details.

Do not send authentication tokens, passwords or entire sources. Privacy requests can also go to sales@goonies.kr.