Each Agent Run Now Works In Its Own Copy Of Your Code

Summary

The single most common reason an agent run failed to even start was a leftover file from a previous run sitting in the shared folder it worked in. Agent runs on the same machine no longer share that folder: each run now gets its own isolated, disposable copy of your project's code, used only for that run and cleaned up automatically when it ends.

What You Gain

Previously, every agent run on a given machine worked inside the same local copy of your project. If one run left anything behind, every later run on that machine could be refused outright, simply because the folder wasn't in the clean state expected. The fix usually meant someone manually clearing out that shared folder.

Now each run works in its own private copy, created fresh before anything starts and discarded the moment the run ends, however it ends. Your project's actual folder is never touched directly. Three agents working on three different tasks at the same time now each work entirely inside their own copy, so nothing one agent does can collide with another.

How It Works

• Every run gets a fresh, disposable copy of your project's code, removed automatically once the run ends.
• You can save an optional setup script per project that prepares a fresh copy before the agent begins working; without one, the agent installs what it needs itself.
• Unfinished work is reported honestly. Work left uncommitted is discarded with the copy; committed work is kept.
• A run that can't get the access it needs fails immediately with a clear reason.
• Code reviews work the same way, examining the pull request in their own disposable copy.

Where to Find It

This applies automatically to every project and every agent run — there's nothing to turn on. An optional setup script can be added from a project's Resources settings.

See what else shipped that month on the Effigate Command Center changelog, or try Effigate Command Center with your own team.