Add Coverage Agent

[HarryCampion]

Adds a custom GitHub Agent for code coverage.

[tschm]

I don't think that's the best place. In general we would like to test files in the src directory. But this has no src directory.

[HarryCampion]

This should be in draft. It's a WIP.

[HarryCampion]

This should be in draft. It's a WIP.

There are a couple patterns that will work that I've investigated.

1. Periodic workflow to check coverage, create a issue and then assign to copilot or another assignee if desired
2. Same workflow put on PR to main

Above will be rhiza config. If no src then no coverage to run however if rhiza itself then can maybe use an agent to see if we are testing the framework properly

Other way is within rhiza-tools, it will run the gh agent-task or allow a custom open ai key, if in agent mode. If no AI is desired it will just create stubs where coverage is low.

I prefer the former as more workflow driven, and rhiza driven

[HarryCampion]

This should be in draft. It's a WIP.

There are a couple patterns that will work that I've investigated.

1. Periodic workflow to check coverage, create a issue and then assign to copilot or another assignee if desired
2. Same workflow put on PR to main

Above will be rhiza config. If no src then no coverage to run however if rhiza itself then can maybe use an agent to see if we are testing the framework properly

Other way is within rhiza-tools, it will run the gh agent-task or allow a custom open ai key, if in agent mode. If no AI is desired it will just create stubs where coverage is low.

I prefer the former as more workflow driven, and rhiza driven

Would like to hear your thoughts @tschm

[tschm]

I have no firsthand experience with agents. The goal should be to keep it as simple as possible for the user. Also let's not hit comm the copilot with every push