Random testing results in duplicate testing (wasting tokens / time)

[azizi66]

there needs to be a counter in the database that counts number of passes and apply the random testing to remaining once

otherwise there would be duplicates and testing takes forever

https://github.com/leonvanzyl/autocoder/blob/122f03dc218c544c78d21547ee84b83da3326037/mcp_server/feature_mcp.py#L173

[fnpanic]

I like the random testing because AI sometimes brakes things which worked. This is way more efficent then a full regression test of all features.

Maybe "random" is not ideal but a good start.