Repository navigation
Feature Proposal: Automatic Task Resumption After Usage Limit Reset #61614
ElieAngwandi
started this conversation in
Ideas
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Product: OpenAI Codex
Feature name: Scheduled Task Resumption
Suggested capability: Auto-Resume on Quota Reset
I would like to propose a feature that allows Codex users to automatically resume interrupted tasks when their usage quota resets.
When working on complex software engineering tasks, Codex may spend considerable time analyzing codebases, implementing features, refactoring components, running tests, and debugging issues. However, when a usage limit is reached, the execution may stop before the task is completed.
Currently, users may need to manually return to the interrupted task, review its progress, and instruct Codex to continue.
A built-in scheduling mechanism could eliminate this friction by allowing users to schedule automatic task resumption as soon as their usage quota becomes available again.
The goal is simple:
Let Codex continue working on a task after a quota interruption without requiring the user to manually restart the workflow.
Usage limits can interrupt otherwise productive autonomous coding sessions.
This creates several problems:
This becomes increasingly relevant as developers delegate more complex tasks to Codex.
Introduce an Automatic Task Resumption option whenever a task is interrupted due to a usage limit.
When Codex detects that the task cannot continue because the user’s quota has been exhausted, it should offer the following options:
If automatic resumption is selected, Codex would preserve the task’s execution state and automatically restart the workflow once sufficient usage capacity becomes available.
The user should not need to repeat the original prompt or manually reconstruct the previous context.
When a task is interrupted, Codex could display a notification such as:
Task interrupted – Usage limit reached
Your task has been paused because your usage limit has been reached.
Would you like Codex to automatically resume this task when your usage limit resets?
Once scheduled, the task would display a status such as:
Scheduled for automatic resumption
Codex would then notify the user when execution resumes and when the task is completed.
A. Automatic quota detection
Detect when the user’s usage capacity becomes available again and trigger the scheduled task.
B. Execution state preservation
Maintain a structured checkpoint containing:
C. Context-aware resumption
Before continuing, Codex should restore the previous execution context, inspect the current workspace, and verify that previously completed operations do not need to be repeated.
D. Task scheduling and queue management
Allow users to schedule multiple interrupted tasks and manage them through a dedicated queue.
Potential controls include:
E. Notifications
Notify users when:
Beyond basic automatic resumption, this feature could evolve into a broader task-continuity system.
Priority-based execution: When multiple tasks are waiting, Codex could resume them according to user-defined priorities.
Execution policies: Users could define whether Codex should resume only one task at a time or process multiple tasks when capacity allows.
Autonomous continuation: After resuming, Codex could continue through the remaining steps of the original execution plan, including testing and verification, until completion or until another constraint requires intervention.
Resource-aware scheduling: The system could account for available usage capacity and avoid immediately exhausting the newly restored quota on low-priority tasks.
A reliable implementation would require:
Importantly, automatic resumption should respect existing usage limits and plan restrictions. It would not bypass quotas or grant additional usage; it would simply make better use of the capacity available after a reset.
Operations requiring explicit user authorization, such as deployments or destructive actions, should continue to respect existing approval mechanisms.
Area Expected improvement
User experience Less manual intervention
Productivity Better continuity for long-running tasks
Reliability Reduced risk of duplicated work
Autonomous coding More complete end-to-end workflows
User engagement More effective use of Codex
The first implementation could remain relatively simple:
A task queue, priority management, and advanced execution policies could follow in subsequent iterations.
As AI coding agents become increasingly capable of handling complex, multi-step engineering workflows, continuity of execution becomes just as important as the quality of code generation itself.
A task should not necessarily require manual intervention simply because a temporary usage limit has been reached.
Codex should be able to remember where it stopped, wait for available capacity, and continue where it left off.
I believe Automatic Task Resumption would be a valuable addition to Codex, making autonomous coding workflows more seamless, reliable, and productive.
All reactions