Request to make Bash line output configurable #1267

Closed
opened 2026-02-16 17:30:13 -05:00 by yindo · 4 comments
Owner

Originally created by @static-var on GitHub (Aug 12, 2025).

Originally assigned to: @thdxr on GitHub.

I recently saw this change https://github.com/sst/opencode/commit/6aa157cfe6195effa6f9fea3dc18380a48544489 which would limit bash outputs to 1000 lines.

I feel it could be a problem for people working on Gradle Project.
Gradle, the build system, has a tendency to spew out everything which has its own pros and cons but I am very certain that building a fresh project or running tests on any midsize or above project would be affected by this change.

Assumption: I am under assumption that truncating output would also truncate information which is fed to the LLM/model. If the information fed to the model is unaffected then please feel free to just close this request.

Originally created by @static-var on GitHub (Aug 12, 2025). Originally assigned to: @thdxr on GitHub. I recently saw this change https://github.com/sst/opencode/commit/6aa157cfe6195effa6f9fea3dc18380a48544489 which would limit bash outputs to 1000 lines. I feel it could be a problem for people working on Gradle Project. Gradle, the build system, has a tendency to spew out everything which has its own pros and cons but I am very certain that building a fresh project or running tests on any midsize or above project would be affected by this change. Assumption: I am under assumption that truncating output would also truncate information which is fed to the LLM/model. If the information fed to the model is unaffected then please feel free to just close this request.
yindo closed this issue 2026-02-16 17:30:13 -05:00
Author
Owner

@thdxr commented on GitHub (Aug 12, 2025):

i just tweaked it a bit to be a total character limit which is set to 30_000

we can make this configurable...but LLM context windows really get blown out past this point. Do you feel confident it needs all the output?

@thdxr commented on GitHub (Aug 12, 2025): i just tweaked it a bit to be a total character limit which is set to 30_000 we can make this configurable...but LLM context windows really get blown out past this point. Do you feel confident it needs all the output?
Author
Owner

@github-actions[bot] commented on GitHub (Aug 12, 2025):

👋 This issue might be a duplicate of existing issues. Please check:

  • #1556: Do not output more than 2000 lines in commands and file reads - this issue discusses the same concern about tool calls eating up context window, with specific mention of 2000 line limits vs the 1000 line limit mentioned in your issue

If none of these address your specific case, please let us know how this issue differs.

@github-actions[bot] commented on GitHub (Aug 12, 2025): 👋 This issue might be a duplicate of existing issues. Please check: - #1556: Do not output more than 2000 lines in commands and file reads - this issue discusses the same concern about tool calls eating up context window, with specific mention of 2000 line limits vs the 1000 line limit mentioned in your issue If none of these address your specific case, please let us know how this issue differs.
Author
Owner

@static-var commented on GitHub (Aug 12, 2025):

So I have added these flag by default --quiet --warning-mode none when I am building my projects, which suppresses a lot of "debug" output.
But when tests and builds are failing we would likely need those extra lines since gradle spews things in sequential order and errors are usually at the end.

The check you have put is really useful almost always I'd say. But when we want to debug or resolve things, we might need those extra chars / lines.

@static-var commented on GitHub (Aug 12, 2025): So I have added these flag by default `--quiet --warning-mode none` when I am building my projects, which suppresses a lot of "debug" output. But when tests and builds are failing we would likely need those extra lines since gradle spews things in sequential order and errors are usually at the end. The check you have put is really useful almost always I'd say. But when we want to debug or resolve things, we might need those extra chars / lines.
Author
Owner

@rekram1-node commented on GitHub (Dec 27, 2025):

[automated] Closing due to 90+ days of inactivity. Feel free to reopen if you still need this!

@rekram1-node commented on GitHub (Dec 27, 2025): [automated] Closing due to 90+ days of inactivity. Feel free to reopen if you still need this!
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: anomalyco/opencode#1267