Designing Safe Autofixes for eslint-plugin-repo

Autofix quality is a trust problem. A single unsafe fixer can make users disable an otherwise valuable rule.

Safety-first autofix principles

• Prefer small, deterministic edits.
• Avoid fixers when intent cannot be inferred with high confidence.
• Use suggest when multiple valid fixes exist.

Practical safeguards

• Keep fixer ranges narrow and syntax-aware.
• Preserve comments and formatting boundaries.
• Test fixer output for parse validity and idempotence.

Safe autofixes improve adoption; risky autofixes undermine it.