You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
TODO tasks hang indefinitely and never transition to completed state even after underlying work finishes #22288
When using OpenCode’s TODO feature to track progress on multi‑step operations, the TODO items remain in a pending/hanging state and never get marked as completed or closed. This occurs even after the associated tasks have fully finished and new work has started. The UI continues to show the TODO as active, which makes it difficult to trust the current status of the work and clutters the interface with stale items.
Steps to Reproduce
Invoke OpenCode to perform a task that generates one or more TODO items (e.g., a multi‑file refactor, codebase search, or a long‑running code generation).
Observe the TODO list populate with specific action items.
Allow the underlying operations to finish naturally (wait for terminal output / file changes to confirm completion).
Check the TODO panel.
Expected Behavior
Once the corresponding work for a TODO item is completed, its status should update automatically to completed or closed.
The TODO should be removed from the active list or visually distinguished as finished.
Subsequent new actions should create fresh TODO entries rather than leaving the old ones hanging.
Actual Behavior
TODO items remain stuck in a "pending" or "in‑progress" state.
They do not transition to completed nor do they disappear.
This behavior is consistent across multiple runs and different types of tasks.
Environment
OpenCode Version: [e.g., v1.4.3 on Windows]
Additional Context / Logs
This issue significantly impacts the usability of the TODO feature for tracking real‑time progress. It would be great if the TODO lifecycle could be tied more tightly to the actual completion events of the underlying tasks.
Plugins
No response
OpenCode version
1.4.3
Steps to reproduce
Invoke OpenCode to perform a task that generates one or more TODO items (e.g., a multi‑file refactor, codebase search, or a long‑running code generation).
Observe the TODO list populate with specific action items.
Allow the underlying operations to finish naturally (wait for terminal output / file changes to confirm completion).
The description appears to be AI-generated (overly polished, structured language with formal phrasing). Please rewrite in your own words describing what you actually experienced.
The Environment field inside the Description contains unchanged placeholder text: [e.g., v1.4.3 on Windows] — please replace it with your actual environment details (or remove the field if redundant with the fields below).
Please edit this issue to address the above within 2 hours, or it will be automatically closed.
If you believe this was flagged incorrectly, please let a maintainer know.
Unqualified human here. The issue still persists and the stale todo lists can only be removed by starting a new session. That makes long context windows sort of "useless". I find it epic that a dev who makes opencode (a tool to support people in working with AI/LLM) rejects a bug report simply because it appears to be written with AI assistance.
Reacted by GSTbyMM, Fedor Suchkov, Marquilio, goldentoaste, Victor Domingos, jakabbalu, SolidSlime, bennypea and marmazel
Activity
github-actions commentedon Apr 13, 2026
This issue doesn't fully meet our contributing guidelines.
What needs to be fixed:
[e.g., v1.4.3 on Windows]— please replace it with your actual environment details (or remove the field if redundant with the fields below).Please edit this issue to address the above within 2 hours, or it will be automatically closed.
If you believe this was flagged incorrectly, please let a maintainer know.
github-actions commentedon Apr 13, 2026
This issue has been automatically closed because it was not updated to meet our contributing guidelines within the 2-hour window.
Feel free to open a new issue that follows our issue templates.
Exagone313 commentedon May 5, 2026
@GSTbyMM care to create a new issue with a passed version?
perf(local-llm): tune Gemma-4 sampling + add SWA cache-reuse fix + co…
viloocity commentedon Aug 6, 2026
Unqualified human here. The issue still persists and the stale todo lists can only be removed by starting a new session. That makes long context windows sort of "useless". I find it epic that a dev who makes opencode (a tool to support people in working with AI/LLM) rejects a bug report simply because it appears to be written with AI assistance.