From 677e4fb6daf8744a299f2a60546c42b31a5338f3 Mon Sep 17 00:00:00 2001
From: Jack Felke <jrinnfelke@gmail.com>
Date: Tue, 3 Mar 2026 19:44:32 -0700
Subject: [PATCH] fix: scope scoring bug in prompt_score + add 11 unit tests
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- Export scorePrompt for testability
- Fix: long prompts no longer get full scope marks (25→20) without
  actual scope-bounding keywords like 'only', 'just', 'specific'
- Add tests/tools/prompt-score.test.ts with 11 tests covering all
  scoring dimensions, grade boundaries, and edge cases
- Total test count: 43→54
---
 package-lock.json                |  2 +-
 src/tools/prompt-score.ts        |  8 +++-
 tests/tools/prompt-score.test.ts | 75 ++++++++++++++++++++++++++++++++
 3 files changed, 82 insertions(+), 3 deletions(-)
 create mode 100644 tests/tools/prompt-score.test.ts

diff --git a/package-lock.json b/package-lock.json
index 89ef280..4e5a169 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -29,7 +29,7 @@
         "vitest": "^4.0.18"
       },
       "engines": {
-        "node": ">=18"
+        "node": ">=20"
       }
     },
     "node_modules/@esbuild/aix-ppc64": {
diff --git a/src/tools/prompt-score.ts b/src/tools/prompt-score.ts
index 1cecf01..e73e203 100644
--- a/src/tools/prompt-score.ts
+++ b/src/tools/prompt-score.ts
@@ -40,7 +40,7 @@ interface ScoreResult {
   feedback: string[];
 }
 
-function scorePrompt(text: string): ScoreResult {
+export function scorePrompt(text: string): ScoreResult {
   const feedback: string[] = [];
   let specificity: number;
   let scope: number;
@@ -59,8 +59,12 @@ function scorePrompt(text: string): ScoreResult {
   }
 
   // Scope: bounded task
-  if (/\b(only|just|single|one|specific|this)\b/i.test(text) || text.length > 100) {
+  const hasScopeKeyword = /\b(only|just|single|one|specific|this)\b/i.test(text);
+  if (hasScopeKeyword) {
     scope = 25;
+  } else if (text.length > 100) {
+    // Long prompts suggest detail but don't guarantee bounded scope
+    scope = 20;
   } else if (/\b(all|every|entire|whole)\b/i.test(text)) {
     scope = 10;
     feedback.push("🎯 'All/every' is broad — can you narrow the scope?");
diff --git a/tests/tools/prompt-score.test.ts b/tests/tools/prompt-score.test.ts
new file mode 100644
index 0000000..1097bc9
--- /dev/null
+++ b/tests/tools/prompt-score.test.ts
@@ -0,0 +1,75 @@
+import { describe, it, expect } from "vitest";
+import { scorePrompt } from "../../src/tools/prompt-score.js";
+
+describe("scorePrompt", () => {
+  it("gives high score for a well-formed prompt", () => {
+    const result = scorePrompt(
+      "Rename the `handleSubmit` function in `src/components/Form.tsx` to `onFormSubmit`. Only this one function. It should still pass the existing tests."
+    );
+    expect(result.total).toBeGreaterThanOrEqual(80);
+    expect(result.grade).toMatch(/^[AB]/);
+  });
+
+  it("gives low score for a vague prompt", () => {
+    const result = scorePrompt("make it better");
+    expect(result.total).toBeLessThanOrEqual(40);
+    expect(result.grade).toMatch(/^[DF]/);
+    expect(result.feedback.length).toBeGreaterThan(0);
+  });
+
+  it("rewards file paths for specificity", () => {
+    const withPath = scorePrompt("fix src/lib/utils.ts");
+    const without = scorePrompt("fix the utility code");
+    expect(withPath.specificity).toBeGreaterThan(without.specificity);
+  });
+
+  it("rewards scope-bounding keywords", () => {
+    const bounded = scorePrompt("only update the header component");
+    const unbounded = scorePrompt("update components");
+    expect(bounded.scope).toBeGreaterThan(unbounded.scope);
+  });
+
+  it("penalizes broad scope words", () => {
+    const result = scorePrompt("refactor all the files");
+    expect(result.scope).toBeLessThanOrEqual(10);
+    expect(result.feedback.some(f => f.includes("broad"))).toBe(true);
+  });
+
+  it("rewards action verbs", () => {
+    const specific = scorePrompt("extract the validation logic into a helper");
+    const vague = scorePrompt("clean up the validation stuff");
+    expect(specific.actionability).toBeGreaterThan(vague.actionability);
+  });
+
+  it("rewards done conditions", () => {
+    const withDone = scorePrompt("add a test that should return 404 for missing users");
+    const without = scorePrompt("add a test for missing users");
+    expect(withDone.doneCondition).toBeGreaterThan(without.doneCondition);
+  });
+
+  it("treats questions as having implicit done condition", () => {
+    const result = scorePrompt("Why does the login page crash on mobile?");
+    expect(result.doneCondition).toBe(20);
+  });
+
+  it("long prompts get partial scope credit, not full", () => {
+    const longPrompt = "do something with " + "a".repeat(200);
+    const result = scorePrompt(longPrompt);
+    // Should get 20 (partial) not 25 (full) since no scope keywords
+    expect(result.scope).toBe(20);
+  });
+
+  it("returns grade F for minimal input", () => {
+    const result = scorePrompt("hi");
+    expect(result.grade).toBe("F");
+  });
+
+  it("returns congratulatory feedback for perfect prompts", () => {
+    const result = scorePrompt(
+      "In `src/auth/login.ts`, rename only the `validateToken` function to `verifyAuthToken`. The existing test in `tests/auth.test.ts` should still pass."
+    );
+    if (result.total >= 90) {
+      expect(result.feedback.some(f => f.includes("Excellent"))).toBe(true);
+    }
+  });
+});