Troubleshooting

This page focuses on the three most frequent issues first.
More cases can be continuously added to the issue collection table.

1. What to do when login fails?

Typical symptoms

• Connect action fails, or API version probe fails.
• Server URL opens fine in browser, but app login still fails.

Confirm these four key items first

1. Use root server URL: for example https://your-memos.example.com, do not append /api/....
2. Login mode matches credential: Password mode uses username + password; Token mode uses PAT.
3. API version is correct: login page requires selecting server version (0.21.0~0.26.0).
4. Network and certificates: check proxy, firewall, certificate, and system time.

Check by mode

Token mode

• Values pasted as Bearer ... are supported; prefix and spaces are auto-cleaned.
• 401 usually means invalid token or insufficient permission.
• "Version probe failed" is commonly version-selection mismatch first.

Password mode

• Username and password must not be empty.
• If credentials are correct but still fails, check selected server version and network reachability first.
• For 0.23.0, full probe is skipped to reduce login wait time (intentional implementation strategy).

Recommended sequence

1. Refill fields by comparing with Install and First Connection.
2. Switch API version to actual server version and retry.
3. Switch network (Wi-Fi / cellular) and retry.
4. If still failing, keep exact error text and add it to issue collection.

2. What if AI Summary says "no content to send" or cannot start?

What it actually means

The reworked AI Summary reads memos from the local database, inside the selected time range and visibility scope.
If filtering leaves no eligible memos, or no working Chat model is available, the analysis cannot proceed normally.

Most common causes

1. The time range is too narrow.
2. The required visibility scopes are not selected (public / private / protected).
3. Memo body content is empty, so it is skipped.
4. Data has not synced into the local DB yet.
5. AI settings do not contain a working Chat model.
6. You are using the custom template card, but title / description / prompt is incomplete.

Step-by-step checks

1. Expand the time range from last 3 / 7 days to 30 days or a longer custom range.
2. Include the visibility scopes you actually need instead of only checking private-note habits.
3. Pull to refresh or run manual sync, then retry.
4. Open several memos and confirm the body is not empty.
5. Open AI Settings and confirm there is at least one working Chat model.
6. If runs are working but feel weak, add an Embedding model next.
7. If you use the custom template, confirm title, description, and prompt are all saved.

What if the issue is timeout, network error, or connection failure?

Check these first:

• Is Base URL reachable?
• Is API Key correct?
• Does the model key match the provider-side model?
• Are required Extra Headers missing?
• Is there a certificate, proxy, or local network access problem?

3. Issue collection table (continuous updates)

• Issue collection (GitHub Issue)

For each issue, include at least:

• Reproduction steps
• Current versions (app + Memos server)
• Exact error text
• Whether it reproduces consistently