diff --git a/plugins/config-audit/scanners/lib/scoring.mjs b/plugins/config-audit/scanners/lib/scoring.mjs
index a14c4d8..5e7eb16 100644
--- a/plugins/config-audit/scanners/lib/scoring.mjs
+++ b/plugins/config-audit/scanners/lib/scoring.mjs
@@ -4,6 +4,19 @@
  */
 
 import { gradeFromPassRate, WEIGHTS } from './severity.mjs';
+import { humanizeFinding } from './humanizer.mjs';
+
+/**
+ * One-line plain-language context per overall grade. Used when a scorecard
+ * is rendered with `options.humanized: true`.
+ */
+const GRADE_CONTEXT = {
+  A: 'Healthy setup, only minor polish needed',
+  B: 'Good shape — a few items to address',
+  C: 'Some attention needed',
+  D: 'Several issues — prioritize the urgent ones',
+  F: 'Important issues need attention',
+};
 
 // --- Tier weights for utilization calculation ---
 const TIER_WEIGHTS = { t1: 3, t2: 2, t3: 1, t4: 1 };
@@ -235,14 +248,21 @@ export function scoreByArea(scannerResults) {
 /**
  * Derive top 3 actions from GAP findings (T1 first, then T2).
  * @param {object[]} gapFindings
+ * @param {object} [options]
+ * @param {boolean} [options.humanized=false] - When true, return humanized
+ *   recommendations (looked up via humanizer translations).
  * @returns {string[]}
  */
-export function topActions(gapFindings) {
+export function topActions(gapFindings, options = {}) {
   const tierOrder = ['t1', 't2', 't3', 't4'];
   const sorted = [...gapFindings].sort(
     (a, b) => tierOrder.indexOf(a.category) - tierOrder.indexOf(b.category),
   );
-  return sorted.slice(0, 3).map(f => f.recommendation);
+  const top3 = sorted.slice(0, 3);
+  if (options.humanized) {
+    return top3.map(f => humanizeFinding(f).recommendation);
+  }
+  return top3.map(f => f.recommendation);
 }
 
 /**
@@ -307,22 +327,39 @@ export function generateScorecard(areaScores, utilization, maturity, segment, ac
  * Shows only the quality areas (currently 8) — no utilization, maturity, or segment.
  * @param {{ areas: Array<{ name: string, grade: string, score: number }>, overallGrade: string }} areaScores
  * @param {number} opportunityCount - Number of GAP findings (shown as opportunity count)
+ * @param {object} [options]
+ * @param {boolean} [options.humanized=false] - When true, render with plain-language
+ *   grade context and friendlier opportunity phrasing. When false (default),
+ *   render the v5.0.0 verbatim scorecard (backwards-compatible).
  * @returns {string}
  */
-export function generateHealthScorecard(areaScores, opportunityCount) {
+export function generateHealthScorecard(areaScores, opportunityCount, options = {}) {
   const qualityAreas = areaScores.areas.filter(a => a.name !== 'Feature Coverage');
   const avgScore = qualityAreas.length > 0
     ? Math.round(qualityAreas.reduce((s, a) => s + a.score, 0) / qualityAreas.length)
     : 0;
 
+  const humanized = options.humanized === true;
+
   const lines = [];
   lines.push('━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━');
-  lines.push(' Config-Audit Health Score');
+  lines.push(humanized ? ' Configuration health' : ' Config-Audit Health Score');
   lines.push('━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━');
   lines.push('');
-  lines.push(` Health: ${areaScores.overallGrade} (${avgScore}/100)    ${qualityAreas.length} areas scanned`);
+
+  if (humanized) {
+    const context = GRADE_CONTEXT[areaScores.overallGrade] || '';
+    const headline = context
+      ? ` Health: ${areaScores.overallGrade} (${avgScore}/100) — ${context}`
+      : ` Health: ${areaScores.overallGrade} (${avgScore}/100)`;
+    lines.push(headline);
+    lines.push(` ${qualityAreas.length} areas reviewed`);
+  } else {
+    lines.push(` Health: ${areaScores.overallGrade} (${avgScore}/100)    ${qualityAreas.length} areas scanned`);
+  }
+
   lines.push('');
-  lines.push(' Area Scores');
+  lines.push(humanized ? ' Area scores' : ' Area Scores');
   lines.push(' ───────────');
 
   // Format areas in 2-column layout (quality areas only)
@@ -340,7 +377,12 @@ export function generateHealthScorecard(areaScores, opportunityCount) {
 
   if (opportunityCount > 0) {
     lines.push('');
-    lines.push(` ${opportunityCount} ${opportunityCount === 1 ? 'opportunity' : 'opportunities'} available — run /config-audit feature-gap for recommendations`);
+    if (humanized) {
+      const noun = opportunityCount === 1 ? 'way' : 'ways';
+      lines.push(` ${opportunityCount} ${noun} you could get more out of Claude Code — see /config-audit feature-gap`);
+    } else {
+      lines.push(` ${opportunityCount} ${opportunityCount === 1 ? 'opportunity' : 'opportunities'} available — run /config-audit feature-gap for recommendations`);
+    }
   }
 
   lines.push('');
diff --git a/plugins/config-audit/scanners/posture.mjs b/plugins/config-audit/scanners/posture.mjs
index 9482d96..b8ee0c2 100644
--- a/plugins/config-audit/scanners/posture.mjs
+++ b/plugins/config-audit/scanners/posture.mjs
@@ -60,6 +60,7 @@ async function main() {
   let targetPath = '.';
   let outputFile = null;
   let jsonMode = false;
+  let rawMode = false;
   let includeGlobal = false;
   let fullMachine = false;
 
@@ -68,6 +69,8 @@ async function main() {
       outputFile = args[++i];
     } else if (args[i] === '--json') {
       jsonMode = true;
+    } else if (args[i] === '--raw') {
+      rawMode = true;
     } else if (args[i] === '--global') {
       includeGlobal = true;
     } else if (args[i] === '--full-machine') {
@@ -82,14 +85,20 @@ async function main() {
   const filterFixtures = !args.includes('--include-fixtures');
   const result = await runPosture(targetPath, { includeGlobal, fullMachine, filterFixtures });
 
-  if (jsonMode) {
+  // stdout JSON path: --json and --raw both write the v5.0.0-shape result
+  // (byte-identical). Default mode writes nothing to stdout.
+  if (jsonMode || rawMode) {
     const json = JSON.stringify(result, null, 2);
     process.stdout.write(json + '\n');
-  } else {
-    // Terminal scorecard (v3 health format)
+  }
+
+  // stderr scorecard path: --json suppresses; --raw renders v5.0.0 verbatim
+  // (humanized=false); default renders humanized scorecard.
+  if (!jsonMode) {
     const scorecard = generateHealthScorecard(
       { areas: result.areas, overallGrade: result.overallGrade },
       result.opportunityCount,
+      { humanized: !rawMode },
     );
     process.stderr.write('\n' + scorecard + '\n');
   }
diff --git a/plugins/config-audit/tests/lib/scoring-humanizer.test.mjs b/plugins/config-audit/tests/lib/scoring-humanizer.test.mjs
new file mode 100644
index 0000000..739fbca
--- /dev/null
+++ b/plugins/config-audit/tests/lib/scoring-humanizer.test.mjs
@@ -0,0 +1,134 @@
+import { describe, it } from 'node:test';
+import assert from 'node:assert/strict';
+import { generateHealthScorecard, topActions } from '../../scanners/lib/scoring.mjs';
+
+const SAMPLE_AREA_SCORES = {
+  areas: [
+    { id: 'claude_md', name: 'CLAUDE.md', grade: 'A', score: 100, findingCount: 0 },
+    { id: 'settings', name: 'Settings', grade: 'A', score: 90, findingCount: 1 },
+    { id: 'hooks', name: 'Hooks', grade: 'A', score: 100, findingCount: 0 },
+    { id: 'feature_coverage', name: 'Feature Coverage', grade: 'D', score: 30, findingCount: 17 },
+  ],
+  overallGrade: 'A',
+};
+
+const SAMPLE_GAP_FINDINGS = [
+  {
+    id: 'CA-GAP-001',
+    scanner: 'GAP',
+    severity: 'medium',
+    title: 'No CLAUDE.md file',
+    description: 'No project instructions file detected.',
+    recommendation: 'Create a CLAUDE.md file with project-specific guidance.',
+    category: 't1',
+    file: null,
+  },
+  {
+    id: 'CA-GAP-002',
+    scanner: 'GAP',
+    severity: 'medium',
+    title: 'No permissions configured',
+    description: 'No permissions block in settings.',
+    recommendation: 'Add a permissions block to settings.json.',
+    category: 't1',
+    file: null,
+  },
+  {
+    id: 'CA-GAP-003',
+    scanner: 'GAP',
+    severity: 'low',
+    title: 'No status line configured',
+    description: 'No status line.',
+    recommendation: 'Add a status line.',
+    category: 't3',
+    file: null,
+  },
+];
+
+describe('generateHealthScorecard signature change (3-param)', () => {
+  it('2-arg call: backwards-compatible (humanized defaults to false)', () => {
+    const out = generateHealthScorecard(SAMPLE_AREA_SCORES, 17);
+    assert.equal(typeof out, 'string');
+    assert.ok(out.length > 0);
+    assert.ok(out.includes('Config-Audit Health Score'),
+      'non-humanized scorecard should contain v5.0.0 title');
+  });
+
+  it('3-arg call with {humanized: false}: byte-equal to 2-arg call', () => {
+    const twoArg = generateHealthScorecard(SAMPLE_AREA_SCORES, 17);
+    const threeArgFalse = generateHealthScorecard(SAMPLE_AREA_SCORES, 17, { humanized: false });
+    assert.equal(threeArgFalse, twoArg, 'options.humanized=false must produce identical output to 2-arg call');
+  });
+
+  it('3-arg call with {humanized: true}: differs from non-humanized', () => {
+    const nonHumanized = generateHealthScorecard(SAMPLE_AREA_SCORES, 17, { humanized: false });
+    const humanized = generateHealthScorecard(SAMPLE_AREA_SCORES, 17, { humanized: true });
+    assert.notEqual(humanized, nonHumanized,
+      'humanized=true must produce different output from humanized=false');
+  });
+
+  it('3-arg call with {humanized: true}: contains user-friendly phrasing', () => {
+    const humanized = generateHealthScorecard(SAMPLE_AREA_SCORES, 17, { humanized: true });
+    // Must contain at least one humanized cue distinguishing it from v5.0.0 prose
+    const hasGradeContext = /healthy|good shape|attention|polish|setup/i.test(humanized);
+    assert.ok(hasGradeContext,
+      `humanized scorecard must include user-friendly grade context, got:\n${humanized}`);
+  });
+
+  it('preserves area names and scores in both modes', () => {
+    const nonHumanized = generateHealthScorecard(SAMPLE_AREA_SCORES, 17, { humanized: false });
+    const humanized = generateHealthScorecard(SAMPLE_AREA_SCORES, 17, { humanized: true });
+    for (const area of SAMPLE_AREA_SCORES.areas.filter(a => a.name !== 'Feature Coverage')) {
+      assert.ok(nonHumanized.includes(area.name),
+        `non-humanized scorecard must include area name "${area.name}"`);
+      assert.ok(humanized.includes(area.name),
+        `humanized scorecard must include area name "${area.name}"`);
+      assert.ok(nonHumanized.includes(`(${area.score})`),
+        `non-humanized scorecard must include score (${area.score})`);
+      assert.ok(humanized.includes(`(${area.score})`),
+        `humanized scorecard must include score (${area.score})`);
+    }
+  });
+
+  it('opportunity count handling in humanized mode', () => {
+    const humanizedZero = generateHealthScorecard(SAMPLE_AREA_SCORES, 0, { humanized: true });
+    const humanizedMany = generateHealthScorecard(SAMPLE_AREA_SCORES, 17, { humanized: true });
+    assert.ok(humanizedMany.includes('17'), 'humanized scorecard must include opportunity count');
+    // Both paths must remain finite strings
+    assert.equal(typeof humanizedZero, 'string');
+    assert.equal(typeof humanizedMany, 'string');
+  });
+});
+
+describe('topActions humanizer support', () => {
+  it('1-arg call: returns raw recommendations (backwards-compatible)', () => {
+    const actions = topActions(SAMPLE_GAP_FINDINGS);
+    assert.equal(actions.length, 3);
+    assert.equal(actions[0], 'Create a CLAUDE.md file with project-specific guidance.');
+    assert.equal(actions[1], 'Add a permissions block to settings.json.');
+    assert.equal(actions[2], 'Add a status line.');
+  });
+
+  it('2-arg call with {humanized: false}: identical to 1-arg call', () => {
+    const oneArg = topActions(SAMPLE_GAP_FINDINGS);
+    const twoArg = topActions(SAMPLE_GAP_FINDINGS, { humanized: false });
+    assert.deepStrictEqual(twoArg, oneArg);
+  });
+
+  it('2-arg call with {humanized: true}: at least one recommendation differs', () => {
+    const raw = topActions(SAMPLE_GAP_FINDINGS, { humanized: false });
+    const humanized = topActions(SAMPLE_GAP_FINDINGS, { humanized: true });
+    assert.equal(humanized.length, raw.length, 'array length preserved');
+    // The humanizer's GAP TRANSLATIONS replace at least one recommendation (No CLAUDE.md → "Add the file…")
+    const anyDiffer = humanized.some((r, i) => r !== raw[i]);
+    assert.ok(anyDiffer,
+      `humanized=true must change at least one recommendation. raw=${JSON.stringify(raw)} humanized=${JSON.stringify(humanized)}`);
+  });
+
+  it('preserves ordering by tier (t1 → t2 → t3)', () => {
+    const humanized = topActions(SAMPLE_GAP_FINDINGS, { humanized: true });
+    assert.equal(humanized.length, 3);
+    // 1st & 2nd: t1 findings, 3rd: t3 finding (t2 absent in sample)
+    // Both modes preserve this ordering.
+  });
+});
diff --git a/plugins/config-audit/tests/scanners/posture-humanizer.test.mjs b/plugins/config-audit/tests/scanners/posture-humanizer.test.mjs
new file mode 100644
index 0000000..94ecc75
--- /dev/null
+++ b/plugins/config-audit/tests/scanners/posture-humanizer.test.mjs
@@ -0,0 +1,130 @@
+import { describe, it } from 'node:test';
+import assert from 'node:assert/strict';
+import { resolve, dirname } from 'node:path';
+import { fileURLToPath } from 'node:url';
+import { execFile } from 'node:child_process';
+import { promisify } from 'node:util';
+import { readFile, unlink } from 'node:fs/promises';
+
+const exec = promisify(execFile);
+const __dirname = dirname(fileURLToPath(import.meta.url));
+const REPO = resolve(__dirname, '../..');
+const CLI = resolve(REPO, 'scanners/posture.mjs');
+const FIXTURE = resolve(REPO, 'tests/fixtures/marketplace-medium');
+const POSTURE_JSON_SNAPSHOT = resolve(REPO, 'tests/snapshots/v5.0.0/posture.json');
+const POSTURE_STDERR_SNAPSHOT = resolve(REPO, 'tests/snapshots/v5.0.0-stderr/posture.txt');
+
+/**
+ * Normalize a runPosture result for snapshot comparison by zeroing out
+ * time-varying fields and machine-specific paths.
+ */
+function normalizePosture(p) {
+  const out = JSON.parse(JSON.stringify(p));
+  if (out.scannerEnvelope) {
+    if (out.scannerEnvelope.meta) {
+      out.scannerEnvelope.meta.target = '<TARGET>';
+      out.scannerEnvelope.meta.timestamp = '<TIMESTAMP>';
+    }
+    if (Array.isArray(out.scannerEnvelope.scanners)) {
+      for (const s of out.scannerEnvelope.scanners) {
+        s.duration_ms = 0;
+      }
+    }
+  }
+  return out;
+}
+
+/** Strip time-varying durations (Xms) so progress lines compare verbatim across runs. */
+function normalizeStderr(s) {
+  return s.replace(/\(\d+ms\)/g, '(0ms)');
+}
+
+async function runPosture(flags) {
+  const proc = await exec('node', [CLI, FIXTURE, ...flags], {
+    timeout: 60000,
+    cwd: REPO,
+  }).catch(err => err); // posture exits non-zero on findings — capture either way
+  return {
+    stdout: proc.stdout || '',
+    stderr: proc.stderr || '',
+  };
+}
+
+describe('posture humanizer wiring (Step 6)', () => {
+  describe('--json mode (SC-6: byte-equal stdout)', () => {
+    it('stdout JSON deepEquals v5.0.0 snapshot', async () => {
+      const { stdout } = await runPosture(['--json']);
+      const actual = JSON.parse(stdout);
+      const expected = JSON.parse(await readFile(POSTURE_JSON_SNAPSHOT, 'utf-8'));
+      assert.deepStrictEqual(normalizePosture(actual), normalizePosture(expected));
+    });
+
+    it('does NOT write a scorecard to stderr (suppressed)', async () => {
+      const { stderr } = await runPosture(['--json']);
+      assert.ok(!stderr.includes('Config-Audit Health Score'),
+        'stderr must NOT contain scorecard in --json mode');
+      assert.ok(!stderr.includes('Configuration health'),
+        'stderr must NOT contain humanized scorecard in --json mode');
+    });
+
+    it('preserves v5.0.0 finding shape (no humanizer fields in scannerEnvelope)', async () => {
+      const { stdout } = await runPosture(['--json']);
+      const actual = JSON.parse(stdout);
+      for (const s of actual.scannerEnvelope.scanners) {
+        for (const f of s.findings) {
+          assert.equal(f.userImpactCategory, undefined,
+            `${f.id}: --json findings must not have userImpactCategory`);
+        }
+      }
+    });
+  });
+
+  describe('--raw mode (SC-7: byte-equal stdout + verbatim stderr)', () => {
+    it('stdout JSON deepEquals v5.0.0 snapshot', async () => {
+      const { stdout } = await runPosture(['--raw']);
+      const actual = JSON.parse(stdout);
+      const expected = JSON.parse(await readFile(POSTURE_JSON_SNAPSHOT, 'utf-8'));
+      assert.deepStrictEqual(normalizePosture(actual), normalizePosture(expected));
+    });
+
+    it('stderr scorecard verbatim matches v5.0.0 stderr snapshot', async () => {
+      const { stderr } = await runPosture(['--raw']);
+      const expected = await readFile(POSTURE_STDERR_SNAPSHOT, 'utf-8');
+      // Compare the scorecard portion verbatim (modulo timing in scanner progress lines)
+      assert.equal(normalizeStderr(stderr).trim(), normalizeStderr(expected).trim());
+    });
+
+    it('preserves v5.0.0 finding shape in stdout', async () => {
+      const { stdout } = await runPosture(['--raw']);
+      const actual = JSON.parse(stdout);
+      for (const s of actual.scannerEnvelope.scanners) {
+        for (const f of s.findings) {
+          assert.equal(f.userImpactCategory, undefined,
+            `${f.id}: --raw findings must not have userImpactCategory`);
+        }
+      }
+    });
+  });
+
+  describe('default mode (humanized scorecard)', () => {
+    it('writes humanized scorecard to stderr', async () => {
+      const { stderr } = await runPosture([]);
+      // Humanized scorecard must contain at least one user-friendly cue not in raw v5.0.0
+      const hasGradeContext = /healthy|good shape|attention|polish|setup/i.test(stderr);
+      assert.ok(hasGradeContext,
+        `humanized stderr scorecard must contain user-friendly phrasing, got:\n${stderr}`);
+    });
+
+    it('does NOT write JSON to stdout in default mode', async () => {
+      const { stdout } = await runPosture([]);
+      assert.equal(stdout.trim(), '', 'default mode must not write JSON to stdout');
+    });
+
+    it('humanized scorecard differs byte-wise from v5.0.0 stderr', async () => {
+      const { stderr } = await runPosture([]);
+      const expected = await readFile(POSTURE_STDERR_SNAPSHOT, 'utf-8');
+      assert.notEqual(normalizeStderr(stderr).trim(), normalizeStderr(expected).trim(),
+        'humanized stderr must differ from v5.0.0 verbatim stderr');
+    });
+  });
+});
diff --git a/plugins/config-audit/tests/scanners/posture.test.mjs b/plugins/config-audit/tests/scanners/posture.test.mjs
index 9222727..40d6a4f 100644
--- a/plugins/config-audit/tests/scanners/posture.test.mjs
+++ b/plugins/config-audit/tests/scanners/posture.test.mjs
@@ -92,8 +92,10 @@ describe('posture.mjs CLI — minimal project', () => {
 });
 
 describe('posture.mjs CLI — terminal output (v3 health format)', () => {
-  it('scorecard contains health sections', async () => {
-    const { stderr } = await runPosture([resolve(FIXTURES, 'healthy-project')]);
+  // These assertions verify the v5.0.0 verbatim scorecard prose. Default mode
+  // is humanized as of v5.1.0 (Wave 3); --raw is the explicit v5.0.0 path.
+  it('scorecard contains health sections (v5.0.0 verbatim via --raw)', async () => {
+    const { stderr } = await runPosture([resolve(FIXTURES, 'healthy-project'), '--raw']);
     assert.ok(stderr.includes('Config-Audit Health Score'));
     assert.ok(stderr.includes('Health:'));
     assert.ok(stderr.includes('Area Scores'));
@@ -101,14 +103,14 @@ describe('posture.mjs CLI — terminal output (v3 health format)', () => {
   });
 
   it('scorecard does NOT contain legacy metrics', async () => {
-    const { stderr } = await runPosture([resolve(FIXTURES, 'healthy-project')]);
+    const { stderr } = await runPosture([resolve(FIXTURES, 'healthy-project'), '--raw']);
     assert.ok(!stderr.includes('Maturity:'));
     assert.ok(!stderr.includes('Utilization:'));
     assert.ok(!stderr.includes('Segment:'));
   });
 
   it('scorecard excludes Feature Coverage from area display', async () => {
-    const { stderr } = await runPosture([resolve(FIXTURES, 'healthy-project')]);
+    const { stderr } = await runPosture([resolve(FIXTURES, 'healthy-project'), '--raw']);
     assert.ok(!stderr.includes('Feature Coverage'));
   });
 });