A Stuck Agent Now Tells You, and One Click Restarts It
Stuck agent runs are now easier to spot and fix, with clearer failure reporting, a preserved history of what went wrong, and a one-click retry.
Summary
A sub-task could previously sit untouched for a long time with no one told why, caused by a handful of separate reliability gaps in how Effigate Command Center detects, reports, and recovers from a stuck or failed agent run. All of them are fixed, and failed runs are now easier to spot, understand, and restart.
What You Gain
Previously, a run could be incorrectly treated as blocked by unrelated files an agent's own test suite created inside dependency folders, even though nothing about your actual work had changed. A run could also be marked as failed purely because its activity log was slow to upload, discarding work an agent had already finished and committed. When a run was automatically retried, the record of its original failure was sometimes wiped out, leaving no way to tell what had actually gone wrong the first time. And when a run failed in a way nobody was actively watching — common for work handed entirely to an agent — the notification that should have gone out sometimes went to nobody at all.
All four of these are fixed. Scratch files an agent's own tests create inside dependency folders no longer block a run. A slow or failed log upload now costs you part of the log at worst, never the actual work. Every failed attempt is recorded in the task's own history before anything retries, so you can always see what was tried and what happened. And when there's no one specific to notify, the alert now falls back to your managers and your project's connected Slack channel instead of going nowhere.
How It Works
- Scratch files an agent's own tests create inside dependency or build folders no longer block the next task from starting.
- A slow or failed log upload no longer causes a finished, committed run to be marked as failed.
- Every failed attempt is saved to the task's history, so you can see exactly what was tried and why it failed.
- If there's no one specifically following a task, failure alerts now reach your managers and your project's Slack channel instead of disappearing.
- A failed run now shows a Retry run button directly on the task and in your Decision Queue — no command line needed.
Where to Find It
Reporting and retry improvements apply automatically across your account. Make sure your desktop app or standalone command-line tool is up to date to get the full set of fixes.
See what else shipped that month on the Effigate Command Center changelog, or try Effigate Command Center with your own team.