Which AI coding-agent workflow do you repeat most often?

We’re launching AI Agent Skills Pack tomorrow: ten editable Markdown workflows for recurring engineering tasks. It came from a simple frustration—coding agents can sound confident without inspecting the real target or verifying the result.

Four complete companion workflows are public and MIT licensed, so the method can be inspected before anyone considers the paid pack:

https://github.com/skyestrela/ai...

I’d value direct feedback from people using coding agents:

• Which task do you repeat most often: review, debugging, tests, security, migrations, deployment, refactoring, PR shipping, or incident analysis?

• What evidence would make a reusable agent workflow trustworthy enough for you to adopt?

• Where do current skills or prompts tend to fail in real repositories?

I’ll use concrete answers to improve the workflows rather than inventing features or traction.