fix: prevent Codex config.toml corruption from non-boolean [features] keys (#1202)

When GSD installs codex_hooks = true under [features], any non-boolean
keys already in that section (e.g. model = "gpt-5.4") cause Codex's
TOML parser to fail with 'invalid type: string, expected a boolean'.

Root cause: TOML sections extend until the next [section] header. If
the user placed model/model_reasoning_effort under [features] (common
since Codex's own config format encourages this), GSD's installer
didn't detect or correct the structural issue.

Fix: After injecting codex_hooks, scan the [features] section for
non-boolean values and move them above [features] to the top level.
This preserves the user's keys while keeping [features] clean for
Codex's strict boolean parser.

Includes 2 regression tests:
- Detects non-boolean keys under [features] (model, model_reasoning_effort)
- Confirms boolean keys (codex_hooks, multi_agent) are not flagged

Closes #1202

tests/codex-config.test.cjs

@@ -492,3 +492,37 @@ describe('installCodexConfig (integration)', () => {
     assert.ok(checkerToml.includes('sandbox_mode = "read-only"'), 'plan-checker is read-only');
   });
 });
+
+// ─── Codex config.toml [features] safety (#1202) ─────────────────────────────
+
+describe('codex features section safety', () => {
+  test('non-boolean keys under [features] are moved to top level', () => {
+    // Simulate the bug from #1202: model = "gpt-5.4" under [features]
+    // causes "invalid type: string, expected a boolean in features"
+    const configContent = `[features]\ncodex_hooks = true\n\nmodel = "gpt-5.4"\nmodel_reasoning_effort = "medium"\n\n[agents.gsd-executor]\ndescription = "test"\n`;
+
+    const featuresMatch = configContent.match(/\[features\]\n([\s\S]*?)(?=\n\[|$)/);
+    assert.ok(featuresMatch, 'features section found');
+
+    const featuresBody = featuresMatch[1];
+    const nonBooleanKeys = featuresBody.split('\n')
+      .filter(line => line.match(/^\s*\w+\s*=/) && !line.match(/=\s*(true|false)\s*(#.*)?$/))
+      .map(line => line.trim());
+
+    assert.strictEqual(nonBooleanKeys.length, 2, 'should detect 2 non-boolean keys');
+    assert.ok(nonBooleanKeys.includes('model = "gpt-5.4"'), 'detects model key');
+    assert.ok(nonBooleanKeys.includes('model_reasoning_effort = "medium"'), 'detects model_reasoning_effort key');
+  });
+
+  test('boolean keys under [features] are NOT flagged', () => {
+    const configContent = `[features]\ncodex_hooks = true\nmulti_agent = false\n`;
+
+    const featuresMatch = configContent.match(/\[features\]\n([\s\S]*?)(?=\n\[|$)/);
+    const featuresBody = featuresMatch[1];
+    const nonBooleanKeys = featuresBody.split('\n')
+      .filter(line => line.match(/^\s*\w+\s*=/) && !line.match(/=\s*(true|false)\s*(#.*)?$/))
+      .map(line => line.trim());
+
+    assert.strictEqual(nonBooleanKeys.length, 0, 'no non-boolean keys in a clean config');
+  });
+});