anomalyco/opencode
1
0
Fork 0
mirror of https://github.com/anomalyco/opencode.git synced 2026-07-22 10:15:31 -04:00
Code Issues 5.9k Packages Projects Releases 19 Wiki Activity

Chained prompts executing together instead of one after the other #2047

New Issue
Open
opened 2026-02-16 17:33:55 -05:00 by yindo · 14 comments
yindo commented 2026-02-16 17:33:55 -05:00
Owner
Copy Link
Copy Source

Originally created by @bma-d on GitHub (Oct 11, 2025).

Originally assigned to: @rekram1-node on GitHub.

Exactly as the title is,
huge fan of OpenCode but really miss chaining prompts,
it seems like when the agent reads files or does an action it ignores any other consecutive task and takes the next lined up prompt instead. I've seen this happen more frequently when the next lined up prompt is a command.

Originally created by @bma-d on GitHub (Oct 11, 2025). Originally assigned to: @rekram1-node on GitHub. Exactly as the title is, huge fan of OpenCode but really miss chaining prompts, it seems like when the agent reads files or does an action it ignores any other consecutive task and takes the next lined up prompt instead. I've seen this happen more frequently when the next lined up prompt is a command.
yindo commented 2026-02-16 17:33:55 -05:00
Author
Owner
Copy Link
Copy Source

@github-actions[bot] commented on GitHub (Oct 11, 2025):

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

  • #2609: Similar behavior where queued messages execute together instead of sequentially, specifically when commands like /compact are running

Feel free to ignore if none of these address your specific case.

@github-actions[bot] commented on GitHub (Oct 11, 2025): This issue might be a duplicate of existing issues. Please check: - #2609: Similar behavior where queued messages execute together instead of sequentially, specifically when commands like /compact are running Feel free to ignore if none of these address your specific case.
yindo commented 2026-02-16 17:33:55 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Oct 11, 2025):

does this relate to queuing? I just wanna make sure I understand what you mean by chaining prompts

@rekram1-node commented on GitHub (Oct 11, 2025): does this relate to queuing? I just wanna make sure I understand what you mean by chaining prompts
yindo commented 2026-02-16 17:33:55 -05:00
Author
Owner
Copy Link
Copy Source

@bma-d commented on GitHub (Oct 11, 2025):

does this relate to queuing? I just wanna make sure I understand what you mean by chaining prompts

Yes it's queuing, my bad.

@bma-d commented on GitHub (Oct 11, 2025): > does this relate to queuing? I just wanna make sure I understand what you mean by chaining prompts Yes it's queuing, my bad.
yindo commented 2026-02-16 17:33:55 -05:00
Author
Owner
Copy Link
Copy Source

@goniz commented on GitHub (Oct 11, 2025):

I'm thinking about this as a feature.
Some times I have multiple comments to steer the agent and I want it to see them all for the next turn without having to super engineer the prompt

@goniz commented on GitHub (Oct 11, 2025): I'm thinking about this as a feature. Some times I have multiple comments to steer the agent and I want it to see them all for the next turn without having to super engineer the prompt
yindo commented 2026-02-16 17:33:55 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Oct 11, 2025):

Some times I have multiple comments to steer the agent and I want it to see them all for the next turn without having to super engineer the prompt

Wdym by next turn? Are you saying when you queue they queue too quickly?

@rekram1-node commented on GitHub (Oct 11, 2025): > Some times I have multiple comments to steer the agent and I want it to see them all for the next turn without having to super engineer the prompt Wdym by next turn? Are you saying when you queue they queue too quickly?
yindo commented 2026-02-16 17:33:55 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Oct 11, 2025):

Right now we do the same thing that claude code does

@rekram1-node commented on GitHub (Oct 11, 2025): Right now we do the same thing that claude code does
yindo commented 2026-02-16 17:33:55 -05:00
Author
Owner
Copy Link
Copy Source

@goniz commented on GitHub (Oct 11, 2025):

Some times I have multiple comments to steer the agent and I want it to see them all for the next turn without having to super engineer the prompt

Wdym by next turn? Are you saying when you queue they queue too quickly?

I don't think anything needs changing.

I like it that I'm able to send multiple queued messages while the agent is still working.
When it's done it will receive all queued messages in a single request instead of going through them one by one.

Attaching an "example" conversation: https://opencode.ai/docs/s/xvgdyOQZ/

@goniz commented on GitHub (Oct 11, 2025): > > Some times I have multiple comments to steer the agent and I want it to see them all for the next turn without having to super engineer the prompt > > Wdym by next turn? Are you saying when you queue they queue too quickly? I don't think anything needs changing. I like it that I'm able to send multiple queued messages while the agent is still working. When it's done it will receive all queued messages in a single request instead of going through them one by one. Attaching an "example" conversation: https://opencode.ai/docs/s/xvgdyOQZ/
yindo commented 2026-02-16 17:33:56 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Oct 11, 2025):

Ah sorry I think I misread your comment:

I'm thinking about this as a feature.

my bad

@rekram1-node commented on GitHub (Oct 11, 2025): Ah sorry I think I misread your comment: > I'm thinking about this as a feature. my bad
yindo commented 2026-02-16 17:33:56 -05:00
Author
Owner
Copy Link
Copy Source

@bma-d commented on GitHub (Oct 11, 2025):

Some times I have multiple comments to steer the agent and I want it to see them all for the next turn without having to super engineer the prompt

Wdym by next turn? Are you saying when you queue they queue too quickly?

I don't think anything needs changing.

I like it that I'm able to send multiple queued messages while the agent is still working. When it's done it will receive all queued messages in a single request instead of going through them one by one.

Attaching an "example" conversation: https://opencode.ai/docs/s/xvgdyOQZ/

I think Codex handles this the right way, as in you have the option to edit the previous message using alt + ↑ but also queue.

@bma-d commented on GitHub (Oct 11, 2025): > > > Some times I have multiple comments to steer the agent and I want it to see them all for the next turn without having to super engineer the prompt > > > > > > Wdym by next turn? Are you saying when you queue they queue too quickly? > > I don't think anything needs changing. > > I like it that I'm able to send multiple queued messages while the agent is still working. When it's done it will receive all queued messages in a single request instead of going through them one by one. > > Attaching an "example" conversation: https://opencode.ai/docs/s/xvgdyOQZ/ I think Codex handles this the right way, as in you have the option to edit the previous message using alt + ↑ but also queue.
yindo commented 2026-02-16 17:33:56 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Oct 11, 2025):

Okay I played with codex, opencode intentionally has a different queueing strategy

But I did see the alt + arrow, which allows you to edit queued message, we could add that

@rekram1-node commented on GitHub (Oct 11, 2025): Okay I played with codex, opencode intentionally has a different queueing strategy But I did see the alt + arrow, which allows you to edit queued message, we could add that
yindo commented 2026-02-16 17:33:56 -05:00
Author
Owner
Copy Link
Copy Source

@bma-d commented on GitHub (Oct 20, 2025):

Okay I played with codex, opencode intentionally has a different queueing strategy

But I did see the alt + arrow, which allows you to edit queued message, we could add that

That would be fantastic, I would be primarily using OpenCode if this was added 👍

@bma-d commented on GitHub (Oct 20, 2025): > Okay I played with codex, opencode intentionally has a different queueing strategy > > But I did see the alt + arrow, which allows you to edit queued message, we could add that That would be fantastic, I would be primarily using OpenCode if this was added 👍
yindo commented 2026-02-16 17:33:56 -05:00
Author
Owner
Copy Link
Copy Source

@rekram1-node commented on GitHub (Oct 21, 2025):

Good to know!

We are in the midst of a rewrite of the tui but after that is done (this week) we can readdress the ux around some of these things better

@rekram1-node commented on GitHub (Oct 21, 2025): Good to know! We are in the midst of a rewrite of the tui but after that is done (this week) we can readdress the ux around some of these things better
yindo commented 2026-02-16 17:33:56 -05:00
Author
Owner
Copy Link
Copy Source

@sai-sy commented on GitHub (Nov 24, 2025):

@rekram1-node I'll "chain" prompts and it will forget some of them. even if I just sent one then after a little bit send another, it might fully abandon the first task when it get's a break from it. gpt-5-codex haven't tested on gpt-5.1-codex

@sai-sy commented on GitHub (Nov 24, 2025): @rekram1-node I'll "chain" prompts and it will forget some of them. even if I just sent one then after a little bit send another, it might fully abandon the first task when it get's a break from it. gpt-5-codex haven't tested on gpt-5.1-codex
yindo commented 2026-02-16 17:33:56 -05:00
Author
Owner
Copy Link
Copy Source

@98caa9deab129460e181aae781d377 commented on GitHub (Dec 1, 2025):

I've been running into this too — queued messages sometimes get lost or conflated. I know models can struggle to keep track of things in longer contexts anyway, but consecutive user messages currently serialize with nothing between them, which might not be helping.
Also feels a bit off that hitting enter to send separate messages doesn't preserve any separation.
Put up a one-liner PR (#4939) that just adds a newline between them. Perhaps a middle ground — keeps the batching but with clearer separation. Small tweak, not sure how much it helps.

@98caa9deab129460e181aae781d377 commented on GitHub (Dec 1, 2025): I've been running into this too — queued messages sometimes get lost or conflated. I know models can struggle to keep track of things in longer contexts anyway, but consecutive user messages currently serialize with nothing between them, which might not be helping. Also feels a bit off that hitting enter to send separate messages doesn't preserve any separation. Put up a one-liner PR ([#4939](https://github.com/sst/opencode/pull/4939)) that just adds a newline between them. Perhaps a middle ground — keeps the batching but with clearer separation. Small tweak, not sure how much it helps.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
acp
bad
bug
discussion
docs
documentation
enhancement
extension
good first issue
growth
help-wanted
mcp
model-problem
needs:compliance
nix
opentui
perf
pull-request
Mirrored from GitHub Pull Request
 tui
web
windows
zen
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: anomalyco/opencode#2047
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?