/hyperpower:usage

Pass rates, fix attempts and failed checks.

Reference

/hyperpower:usage --since 30d --by stage

When to use it

hyperpowers seems slower or less accurate than it was last month.

What you get

Run counts, how often each stage passed, fix rounds per run, check failures, how many assumptions were verified or refuted, and the median run time.

How it works

Shows how your runs are going. --since takes 7d, 30d or all. --by groups the results by stage, agent, model or run.

Watch two numbers:

  1. Check failure rate. How often the build step produces code that fails your own checks.
  2. Fix rounds per task. How many attempts it takes to make those checks pass.

If both are rising, your project rules no longer match your code. The code changed and the rules did not, so hyperpowers is working from an old picture of your repo. Run /hyperpower:init --refresh to fix it.

Also look at how many assumptions were verified compared with refuted. Many refuted assumptions mean it is guessing about your code more than reading it.