OpenClaw's Skill Workshop now does a better job explaining when self-learning is enabled but weekly reviews are paused.
The change landed in PR #142840, titled fix(ui): explain paused weekly skill reviews. It addresses a subtle but important mismatch in the Control UI: users could see Self-learning turned on in Skill Workshop while global cron automation was disabled, leaving weekly reviews paused with no obvious explanation.
That kind of state is easy to misunderstand. If self-learning says it is on, users reasonably expect the scheduled review loop to run. If the scheduler is globally off, the UI needs to explain the difference between the feature setting and the automation engine that makes weekly review happen.
The New Warning
The fix adds a compact warning beside Self-learning when all of the relevant conditions are true:
- Self-learning is in automatic mode, including the default automatic behavior.
- Global cron automation is disabled.
- The saved configuration snapshot is available.
The warning is intentionally narrow. It disappears when cron is enabled, Self-learning is off, Self-learning is proposal-only, or configuration is unavailable. That proposal-only distinction came from review: proposal-only capture is not eligible for weekly reviews, so showing the cron warning there would have been misleading.
The link in the warning opens the cron controls at /settings/automation?section=cron, using the app's route helper so mounted deployments keep their path prefix.
Why This Is More Than Polish
Skill Workshop is becoming one of OpenClaw's durable surfaces. It is where users can shape repeatable behavior, propose reusable skills, and let the system learn from completed work. That makes status clarity matter.
A silent pause is dangerous because it looks like the system is working when it is not. The new warning tells users why weekly reviews are paused and gives them a direct path to the setting that resumes them.
The PR is careful about scope. It does not add polling, scheduler state, backend settings, or a new store. It reuses the existing subscription, warning icon, route helper, and translation catalog. The author describes the production delta as presentation-only: 54 additions and 14 deletions, including 17 CSS lines for the warning.
That restraint is valuable. The UI needed an explanation, not a new automation model.
Evidence From The PR
The PR includes a baseline report: a real Gateway/UI showed Self-learning on with cron off and no explanation. The new regression failed against that baseline, then passed after the warning was added.
Fourteen focused tests passed: nine header cases and five existing Self-learning page cases. The tests cover stale pause text during config refresh and misleading cron advice in proposal-only mode. Formatting, syntax lint, scoped styles, translation validation, and size checks also passed.
The author also reports real browser checks for warning visibility, both relevant settings, keyboard activation, direct navigation to the cron tab, saved-state return, and narrow layout. One limitation is explicitly called out: an isolated type-aware lint attempt exceeded its memory cap, so the PR does not claim that pass locally.
The User-Facing Win
For daily OpenClaw users, this is a small piece of honesty in the interface. Self-learning can be enabled while the scheduled weekly review mechanism is paused, and now the UI says so.
That reduces confusion without making the feature feel broken. The warning does not disable Self-learning or imply all learning has stopped. It simply explains the missing scheduled review loop and points to the exact automation control that can resume it.