OpenCode Go subscription usage doesn't shown in the apps

[rizkyfaza20]

Description

While check the usage and enable plugin for opencode, it seems the status usage doesn't shown as expected

Steps to Reproduce

1. run the apps
2. enable plugin opencode go
3. login via dashboard
4. the preview dashboard was empty
5. reverse the process and check integration, probably from sqlite3, but need more specific case

Expected Behavior

should shown the usage quota 5h, weekly, and monthly

Actual Behavior

output stats were blank

OpenUsage Version

0.6.31

Operating System

macOS (Intel)

Screenshots

No response

Additional Context

OS: Zorin 18

[rizkyfaza20]

openusage-2026-06-18-to-2026-06-18.csv

attached export usage

[symonbaikov]

Thanks for the detailed report and for attaching the export.

I checked the CSV. OpenCode Go is detected with plan Go, but every snapshot only contains Status: No usage data; there are no Session, Weekly, or Monthly rows. That points to OpenUsage failing to read or parse the local OpenCode SQLite history, not just the UI hiding the values.

We’ll investigate this path.

One thing that would help: could you confirm which OS this was captured on? The issue says macOS (Intel), but the additional context says Zorin 18.

Could you also share the OpenUsage app logs around the refresh time, especially any lines mentioning opencode-go, sqlite, or No usage data? Please don’t share auth.json, API keys, or the full opencode.db if it contains private prompts. For now, just confirming whether ~/.local/share/opencode/opencode.db exists on that machine and whether OpenCode Go was used locally before the export would already help.

Thanks again. The export was useful and narrowed this down.

[rizkyfaza20]

Hi Symon,

I'm using Zorin 18 in current, since i can't find any listed OS for Linux so I put additional context

Yeah sure, i also want to put the logs, and I also current debug and check what happened in background, if i put the change to OpenUsage, is it okay?

thanks

[rizkyfaza20]

*i mean to contribute the fix if can

[symonbaikov]

Yes, absolutely. Contributions are welcome!

Please feel free to investigate and open a PR if you find the fix. The most useful starting point is the OpenCode Go plugin path around local SQLite history reading.

Please avoid including any private data in the PR or issue: no API keys, no auth.json, and no full opencode.db if it contains prompts or account data. Logs are useful, but please redact anything sensitive before sharing.

If you open a PR, please link it to this issue so we can review it in context. Thanks for digging into this.

[rizkyfaza20]

hi @symonbaikov

have been digging up the issue, hopefully this proposal can be check with you, i have tried to test up in my local, it seems good, but there slights of things probably need to check

• while i try to refresh, it seems the usage that in opencode.db take a different approach, so i reproduce the process by how if the process were proceed refresh around 1 minute
• some function while to fetch from sqlite; so opencode store all sessions usage in database, and refresh after usage, i improved it, details in the PR

i attached last log of usage in this reproduce process, let me know if you have any feedbacks

opencode-usage.log
openusage-2026-06-18-to-2026-06-22.csv

[symonbaikov]

Hi @rizkyfaza20, thanks for the detailed investigation and the logs that context really helps. The overal change looks prety solid. Hera are a few things to think of:

1. Can you confirm session.time_updated is stored in milliseconds? The plugin code now treats it as ms, but if OpenCode stores it as seconds all time windows will shift to 1970.

2. Could you share a small schema/sample from your opencode.db? e.g. PRAGMA table_info(session); and one or two sanitized rows. That will let us double-check the model JSON shape and the cost/time_updated columns.

3. Please update docs/providers/opencode-go.md — it still shows the old message/data SQL.

4. Add regression tests for the fractional-cost truncation bug (CAST(cost AS TEXT)) and the visibilitychange refresh behavior.
   Once those are in, I’ll re-review and we can get this merged.

[rizkyfaza20]

/@symonbaikov i would like to give an update regarding to your questions:

1. the time session from the local database confirms that session.time_updated is stored in milliseconds:

sqlite3 ~/.local/share/opencode/opencode.db "PRAGMA table_info(session);"

--- Result
14|time_created|INTEGER|1||0
15|time_updated|INTEGER|1||0
21|model|TEXT|0||0
22|cost|REAL|1|0|0

---
sqlite3 ~/.local/share/opencode/opencode.db "
SELECT time_updated, datetime(time_updated/1000, 'unixepoch') AS utc
FROM session WHERE json_valid(model)
AND json_extract(model, '$.providerID') = 'opencode-go' AND cost > 0
LIMIT 2;
"
--- Result

1782176864153|2026-06-23 01:07:44
1782136949911|2026-06-22 14:02:29

it shown that the table_info was printout for use the integer and also time using ms format

2. rows result for sanitize with JSON format

sqlite3 ~/.local/share/opencode/opencode.db "
SELECT id, time_updated, time_created, cost, model
FROM session WHERE json_valid(model)
AND json_extract(model, '$.providerID') = 'opencode-go' AND cost > 0
ORDER BY time_updated DESC LIMIT 2;
"
--- Result

ses_11108a...|1782176864153|1782125649689|3.3981835568|{"id":"deepseek-v4-flash","providerID":"opencode-go","variant":"high"}
ses_1105d8...|1782136949911|1782136861667|0.0072242352|{"id":"deepseek-v4-flash","providerID":"opencode-go","variant":"medium"}

for the PR, i have update the documentation and also regression test in #18 let me know if you to confirm with anything :)

query1.log
query2.log