ck:autoresearch

Autonomous test coverage improvement is achieved by applying an iterative optimization loop to a git-tracked codebase, making atomic code changes and verifying a numeric metric to decide whether to keep or revert each commit.

You can automatically reduce bundle size by running sequential iterations that make focused edits within a declared file scope, verify the size metric, and revert changes that do not meet a configurable minimum delta threshold.

You need a clean git working tree, a declared file scope glob, and a Verify shell command that prints a single numeric value within 30 seconds to enable the automated loop to evaluate each iteration.

Iterative optimization handles regressions by using optional Guard commands that make guard-scoped files read-only, triggering automatic reversion of commits if the guard checks fail or the metric does not improve.

The optimization loop works with lint error reduction by running a Verify command that outputs the error count, iteratively editing scoped files, and committing only changes that lower the numeric lint error metric.

You should not use autonomous iterative code changes if your git working tree is not clean, your metric cannot be printed as a single numeric value by a shell command, or verification takes longer than 30 seconds.