[codex] test(ci): cover post-merge review follow-ups

.github/workflows/qwen-code-pr-review.yml

@@ -355,7 +355,7 @@ jobs:
                 fi
               done
           git worktree prune -v || true
-          echo "stale review worktrees cleaned"
+          echo "stale agent state cleaned"
 
       # SECURITY: checkout trusted base code; /review fetches PR diff context.
       - name: 'Checkout base branch'

packages/core/src/config/config-session-env.test.ts

@@ -108,7 +108,7 @@ const baseParams: ConfigParameters = {
 // flag resets). That cold transform+evaluate runs several seconds and, under
 // a contended CI runner, crosses the 5s default — a flaky timeout, not a hang.
 // The reset is load-bearing for what these tests check, so give them headroom.
-vi.setConfig({ testTimeout: 30_000 });
+vi.setConfig({ testTimeout: 30_000, hookTimeout: 30_000 });
 
 describe('Config sessionEnvClaimed guard', () => {
   let originalEnv: string | undefined;

packages/core/src/skills/skill-activation.test.ts

@@ -5,20 +5,14 @@
  */
 
 import * as path from 'node:path';
-import { describe, expect, it, vi } from 'vitest';
+import { describe, expect, it } from 'vitest';
 import {
   SkillActivationRegistry,
   resolveProjectRelativePath,
   splitConditionalSkills,
 } from './skill-activation.js';
 import type { SkillConfig } from './types.js';
 
-// The integration tests below `await import('../core/coreToolScheduler.js')`
-// just to reach one pure helper, but that drags in the whole scheduler module
-// graph cold. The first such import runs a few seconds and, under a contended
-// CI runner, crosses the 5s default — a flaky timeout, not a hang.
-vi.setConfig({ testTimeout: 30_000 });
-
 function makeSkill(overrides: Partial<SkillConfig>): SkillConfig {
   return {
     name: overrides.name ?? 'test-skill',
@@ -248,6 +242,10 @@ describe('resolveProjectRelativePath', () => {
 });
 
 describe('extractToolFilePaths → SkillActivationRegistry integration', () => {
+  // These tests `await import('../core/coreToolScheduler.js')` just to reach
+  // one pure helper, but that drags in the whole scheduler module graph cold.
+  // Under a contended CI runner, that can cross the 5s default timeout.
+
   // Regression: feed the real candidate output for a `glob` call into
   // the registry and assert end-to-end activation. The earlier per-field
   // extraction (path + pattern as separate candidates) silently failed
@@ -272,7 +270,7 @@ describe('extractToolFilePaths → SkillActivationRegistry integration', () => {
       for (const n of reg.matchAndConsume(c)) activated.add(n);
     }
     expect(Array.from(activated)).toEqual(['tsx-helper']);
-  });
+  }, 30_000);
 
   it('does NOT activate from external glob.path (project-root guard wins)', async () => {
     const { extractToolFilePaths } = await import(
@@ -291,5 +289,5 @@ describe('extractToolFilePaths → SkillActivationRegistry integration', () => {
       for (const n of reg.matchAndConsume(c)) activated.add(n);
     }
     expect(activated.size).toBe(0);
-  });
+  }, 30_000);
 });

scripts/tests/qwen-resolve-workflow.test.js

@@ -14,6 +14,31 @@ const repoRoot = path.resolve(
   '../..',
 );
 
+function escapeRegExp(value) {
+  return value.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+}
+
+function job(workflow, name) {
+  const start = workflow.indexOf(`\n  ${name}:`);
+  if (start === -1) {
+    return '';
+  }
+  const nextJob = workflow.slice(start + 1).search(/\n {2}\S/);
+  return nextJob === -1
+    ? workflow.slice(start)
+    : workflow.slice(start, start + 1 + nextJob);
+}
+
+function step(section, name) {
+  const escaped = escapeRegExp(name);
+  const match = section.match(
+    new RegExp(
+      `\\n\\s+- name:\\s*(['"])${escaped}\\1[\\s\\S]*?(?=\\n\\s+- name:\\s*['"]|\\n\\s{2}[a-zA-Z0-9_-]+:|$)`,
+    ),
+  );
+  return match?.[0] ?? '';
+}
+
 describe('qwen resolve workflow', () => {
   const workflow = readFileSync(
     path.join(repoRoot, '.github/workflows/qwen-code-pr-review.yml'),
@@ -88,6 +113,18 @@ describe('qwen resolve workflow', () => {
     expect(workflow).not.toContain('qwen-fix-conflicts');
   });
 
+  it('isolates review agent state per run', () => {
+    const cleanStep = step(reviewJob, 'Clean stale agent state');
+    const agentStep = step(reviewJob, 'Run review');
+
+    expect(cleanStep).toContain('QWEN_HOME="${RUNNER_TEMP:?}/qwen-home"');
+    expect(cleanStep).toContain('rm -rf "$QWEN_HOME"');
+    expect(cleanStep).toContain('mkdir -p "$QWEN_HOME"');
+    expect(cleanStep).toContain('rm -f /tmp/stage-*.md');
+    expect(cleanStep).toContain('echo "stale agent state cleaned"');
+    expect(agentStep).toContain("QWEN_HOME: '${{ runner.temp }}/qwen-home'");
+  });
+
   // Whole-file `toContain` cannot tell which job a guard lives on. Slice the
   // resolve-pr job so these assertions fail if a future edit drops a guard
   // specifically from the credentialed conflict-resolution path. Bound the slice
@@ -101,6 +138,7 @@ describe('qwen resolve workflow', () => {
     nextJob === -1
       ? workflow.slice(resolveJobStart)
       : workflow.slice(resolveJobStart, resolveJobStart + 1 + nextJob);
+  const reviewJob = job(workflow, 'review-pr');
 
   it('keeps the authorization and scope guards on resolve-pr', () => {
     // /resolve must require write+ permission before any credentialed push.

scripts/tests/qwen-triage-workflow.test.js

@@ -54,6 +54,18 @@ describe('qwen-triage tmux workflow', () => {
     expect(runStep).toContain('"OPENAI_MODEL=$OPENAI_MODEL"');
   });
 
+  it('isolates agent state per run', () => {
+    const cleanStep = step('Clean stale agent state');
+    const runStep = step('Run Qwen Triage');
+
+    expect(cleanStep).toContain('QWEN_HOME="${RUNNER_TEMP:?}/qwen-home"');
+    expect(cleanStep).toContain('rm -rf "$QWEN_HOME"');
+    expect(cleanStep).toContain('mkdir -p "$QWEN_HOME"');
+    expect(cleanStep).toContain('rm -f /tmp/stage-*.md');
+    expect(cleanStep).toContain('echo "stale agent state cleaned"');
+    expect(runStep).toContain("QWEN_HOME: '${{ runner.temp }}/qwen-home'");
+  });
+
   it('reports timeout and infra-error without claiming the flow was exercised', () => {
     const postStep = step('Post tmux result comment');