Skip to content
← All docsRemediation

Explanations and repair prompts

Turn each gap into an explanation, ordered fix steps, and a prompt for your coding agent.

norma fix reuses the same analysis as norma check. It scores the project, then for every requirement that is not met it produces an explanation of why, an ordered list of fix recommendations, and a repair prompt you can paste straight into Claude Code, Cursor, Windsurf, GitHub Copilot, or Bolt.

Remediation runs after scoring and never changes the score. A requirement that is already met is skipped.

Commands

norma fix            explanation, recommended changes, and a repair prompt per finding
norma fix --prompt   output only the repair prompts, one per finding
norma fix --json     the full remediation report as JSON

How a fix is built

The model supplies the reasoning (the explanation and the ordered fix steps); Norma assembles the repair prompt deterministically from those parts, so the wording you paste into your agent is controlled by Norma, not free-form model output. When no provider is available the fix falls back to a template, so norma fix always returns something useful.

Reports that include remediation also stamp a remediationPromptVersion, alongside the judge's promptVersion, so a generated fix is as reproducible as the score it came from.

In CI, set the Action input remediate: true and the PR comment carries the explanation and the fix, not just the score. See GitHub Action.

Did the build keep its promises?

AI can write the code.
Who verifies the intent?