How good is your Cursor rule?

How it is scored

Your rule is read by a large language model alongside a rubric distilled from current best practices for Cursor rules. Each of the seven criteria below is scored 0–10, with feedback that quotes your actual file. The overall 0–100 score is a holistic judgment, not an average, so a single disqualifying problem, like a leaked secret, drags it down hard.

• Scoping & activation

  The frontmatter decides when a rule loads. A clear description, targeted globs, and alwaysApply reserved for genuinely global rules score high. Missing scoping, or alwaysApply on everything, scores low.

• Focused & concise

  A rule is injected into context when it activates, so every line costs tokens. Focused, single-responsibility rules score high; bloated kitchen-sink rules that should be split score low.

• Documents the non-derivable

  The rule should hold what an agent can’t learn from the code: conventions, gotchas, and the “why” behind decisions. Restating file trees or code structure loses points, it wastes tokens and goes stale.

• Specific & actionable

  “Use the cn() helper for conditional classes” beats “write clean components.” Vague aspirations the agent can’t act on score low.

• Examples & references

  Cursor rules are far more effective with a concrete code example of the desired pattern and @file references to canonical files to imitate. Abstract rules with neither score low.

• Structure & scannability

  Clear headings and grouped bullets, with emphasis like ALWAYS / NEVER reserved for the few rules where deviation is costly.

• Guardrails & etiquette

  Explicit boundaries: what must never be touched, branch and PR conventions, destructive-action warnings. Secrets in the file are heavily penalized.