fix: address codex review findings on #168-#172 range

CHANGELOG.md

@@ -48,6 +48,21 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- **Post-commit removal swept the remaining active docs** (codex review of
+  #167): `codeindex init`'s final output no longer promises "Auto-update
+  README_AI.md on commit"; `README_zh.md` (feature bullet, workflow block,
+  plugin-skill table row — the English README was already clean),
+  `examples/ai-integration-guide.md` (10+ mentions incl. a whole
+  Post-commit section and doc-only scenario), and
+  `docs/development/team-workflow-guide.md` ("Three hooks" → two) no
+  longer advertise the removed workflow.
+- **`hooks status` survives an unreadable leftover hook** (codex review):
+  the leftover check's `read_text` now catches `OSError` instead of
+  crashing the informational status output.
+- **JS-test detection prunes venvs and shares the walk skip-set** (codex
+  review of #165): `_has_js_test_files` no longer walks `.venv`/`venv`/
+  `.tox`/`.eggs`, and the skip list is one shared constant with
+  `detect_languages` so the two walkers cannot drift.
 - **Leftover post-commit hooks surfaced** (GH #167 follow-up). A
   `.git/hooks/post-commit` installed by codeindex < 0.37 survives the
   upgrade silently — the dead wrapper's errors go to

README_zh.md

@@ -42,7 +42,6 @@ codeindex 是一个**开源 CLI**，通过**两阶段流水线**把任意代码
 - **两阶段文档流水线**（v0.23.0）——阶段 1：通过 SmartWriter 生成结构化 README_AI.md；阶段 2：AI 为每个模块生成一行式功能描述。AI agent 可以浏览 README_AI.md 层级结构并找到正确的模块，**无需 grep**。
 - **智能索引**——分层文档（概览 → 导航 → 详细），为 AI agent 优化，每个文件 ≤10KB（导航索引，而非技术文档——见 [ADR-005](docs/architecture/adr/005-navigation-disclaimer-and-readme-size-cap.md)）
 - **自动 AI 增强**——当配置了 `ai_command` 时，`scan-all` 会自动启用 AI 模块描述。使用 `--no-ai` 退出
-- **自动更新 hook**——可选的 post-commit hook（`codeindex hooks install`）为变更的目录重新生成 README_AI.md。薄包装模式：`pipx upgrade ai-codeindex` 自动更新 hook 逻辑
 
 ### 解析与分析
 
@@ -123,7 +122,6 @@ codeindex 是无状态的解析层；LoomGraph 是自包含的知识图谱（SQL
 # 个人开发者工作流
 codeindex init                    # 配置 CLAUDE.md 集成
 codeindex scan-all                # 结构化 + AI 描述（自动）
-codeindex hooks install post-commit  # 提交时自动更新
 # Claude Code 读取 README_AI.md → 理解模块用途 → 用 Serena 查细节
 ```
 
@@ -270,7 +268,6 @@ codeindex affected --since HEAD~5
 |-------|--------------|
 | `codeindex:arch` | 从 `README_AI.md` 回答架构 / "X 在哪里" 类问题 |
 | `codeindex:index` | 引导你完成 `codeindex init` → `scan-all` |
-| `codeindex:hooks` | 配置自动更新的 post-commit hook |
 | `codeindex:update-guide` | 刷新你项目 `CLAUDE.md` 中的 codeindex 章节 |
 
 `codeindex init` 还会把 codeindex 章节注入到你项目的 `CLAUDE.md`，

docs/development/team-workflow-guide.md

@@ -152,7 +152,7 @@ They run in CI but are excluded from the fast loop to keep feedback tight.
 
 ## 5. Git Hooks: Automated QA on Every Commit
 
-Three hooks run automatically. You cannot commit or push without passing them.
+Two hooks run automatically. You cannot commit or push without passing them.
 
 ### pre-commit (runs on `git commit`)
 Checks **staged Python files only** — fast, targeted.

examples/ai-integration-guide.md

@@ -10,7 +10,6 @@
 
 **codeindex** is a code indexing tool that generates AI-friendly documentation. It provides **Git Hooks management** for automating:
 - Pre-commit: Code quality checks (lint + debug code detection)
-- Post-commit: Automatic README_AI.md updates
 
 **Your task**: Update the user's project CLAUDE.md or AGENTS.md so future AI sessions understand how Git Hooks work.
 
@@ -75,46 +74,19 @@ ruff check --fix src/
 git commit --no-verify -m "urgent fix"
 ```
 
-#### Post-commit Hook
-
-Runs **after every commit** to keep documentation updated:
-
-1. Analyzes commit changes automatically
-2. Updates `README_AI.md` for affected directories
-3. Creates follow-up commit: `docs: auto-update README_AI.md for <hash>`
-4. Prevents infinite loops (skips if only doc files changed)
-
-**Expected behavior**:
-```bash
-# You commit
-git commit -m "feat: add new feature"
-# Pre-commit checks...
-# ✓ All checks passed!
-
-# Post-commit runs automatically
-# 📝 Post-commit: Analyzing changes...
-# → Updating src/README_AI.md
-# ✓ Auto-commit created
-```
-
-**This is normal!** The auto-commit is expected behavior.
-
 ### For AI Code: Important Behaviors
 
 **When committing code**:
 1. **Hooks run automatically** - No action needed
 2. **If lint errors**: Fix errors before retrying commit
-3. **Post-commit auto-commits**: Expected, don't be alarmed
-4. **Bypass hooks**: Only use `--no-verify` in emergencies
+3. **Bypass hooks**: Only use `--no-verify` in emergencies
 
 **Typical commit flow**:
 ```bash
 git add src/myfile.py
 git commit -m "feat: add feature"
 # → Pre-commit runs (lint + debug check)
 # → If passed: commit succeeds
-# → Post-commit runs (doc update)
-# → May create auto-commit
 ```
 
 **Handling failures**:
@@ -224,19 +196,6 @@ Hooks are installed to `.git/hooks/`:
 - `0` - Pass, commit proceeds
 - `1` - Fail, commit blocked
 
-### Post-commit: Auto Documentation
-
-**Runs**: After commit succeeds
-**Process**:
-1. Analyze commit: `codeindex affected --json`
-2. Get affected directories
-3. Update README_AI.md for each
-4. Create follow-up commit
-
-**Loop Prevention**:
-- Checks if commit only touched doc files
-- Skips if only `README_AI.md` or `PROJECT_INDEX.md` changed
-
 ---
 
 ## 💡 Common Scenarios
@@ -254,11 +213,7 @@ git commit -m "feat: add feature"
 # ✓ Lint passed
 # ✓ No debug code
 
-# Post-commit runs
-# → Updates src/README_AI.md
-# → Creates auto-commit
-
-# Result: 2 commits (original + auto-doc)
+# Result: commit succeeds, docs stay clean
 ```
 
 ### Scenario 2: Lint Errors
@@ -302,20 +257,6 @@ git commit -m "feat: add feature"
 # ✓ Success
 ```
 
-### Scenario 4: Doc-Only Commit
-
-```bash
-# Only update docs
-vim docs/guide.md
-git commit -m "docs: update guide"
-
-# Pre-commit runs (no Python files, skipped)
-# Post-commit runs
-# ✓ Changes below threshold, skipping update
-
-# Result: 1 commit (no auto-doc needed)
-```
-
 ---
 
 ## 🔧 Integration Checklist

src/codeindex/cli_config.py

@@ -79,7 +79,7 @@ def _print_post_init_message():
     console.print("  2. [cyan]codeindex scan-all[/cyan]        → Generate documentation indexes")
     console.print("  3. [cyan]codeindex status[/cyan]           → Check coverage")
     console.print("\n[dim]Optional:[/dim]")
-    console.print("  • [cyan]codeindex hooks install[/cyan]  → Auto-update README_AI.md on commit")
+    console.print("  • [cyan]codeindex hooks install[/cyan]  → Lint hooks (pre-commit / pre-push)")
     console.print(
         "  • [dim]Claude Code:[/dim] [cyan]/plugin install codeindex@codeindex-claude[/cyan]"
     )

src/codeindex/cli_hooks.py

@@ -636,7 +636,10 @@ def status():
         for name in manager.RETIRED_HOOKS:
             hook_path = manager.hooks_dir / name
             if hook_path.exists():
-                content = hook_path.read_text()
+                try:
+                    content = hook_path.read_text()
+                except OSError:
+                    continue  # unreadable — status is informational, don't crash
                 if manager.CODEINDEX_MARKER in content:
                     console.print(
                         f"[yellow]⚠[/yellow] {name}: leftover from a removed "

src/codeindex/init_wizard.py

@@ -163,6 +163,12 @@ def get_parser_install_guidance(languages: list[str]) -> dict:
 
     return result
 
+
+# Directories never worth walking: dependencies / caches / venvs. Shared by
+# detect_languages and _has_js_test_files so one list cannot drift (GH #165).
+_WALK_SKIP_DIRS = {"node_modules", ".git", "__pycache__", ".venv", "venv", ".tox", ".eggs"}
+
+
 def detect_languages(project_dir: Path, max_scan_files: int = 1000) -> List[str]:
     """Auto-detect programming languages in the project.
 
@@ -177,7 +183,7 @@ def detect_languages(project_dir: Path, max_scan_files: int = 1000) -> List[str]
     scanned_files = 0
 
     # Common directories to skip
-    skip_dirs = {"node_modules", ".git", "__pycache__", ".venv", "venv", ".tox", ".eggs"}
+    skip_dirs = _WALK_SKIP_DIRS
 
     for root, dirs, files in os.walk(project_dir):
         # Remove skip directories from walk
@@ -270,10 +276,14 @@ def infer_include_patterns(project_dir: Path) -> List[str]:
 
 
 def _has_js_test_files(project_dir: Path) -> bool:
-    """True if the project contains JS/TS co-located test files or a __tests__ dir."""
+    """True if the project contains JS/TS co-located test files or a __tests__ dir.
+
+    Uncapped walk (early-exit on first hit) is fine here: ``codeindex init``
+    is a one-shot interactive command and a later scan walks the tree anyway.
+    """
     suffixes = (".spec.ts", ".spec.tsx", ".spec.js", ".test.ts", ".test.tsx", ".test.js")
     for root, dirs, files in os.walk(project_dir):
-        dirs[:] = [d for d in dirs if d not in ("node_modules", ".git", "__pycache__")]
+        dirs[:] = [d for d in dirs if d not in _WALK_SKIP_DIRS]
         if "__tests__" in dirs:
             return True
         if any(f.endswith(suffixes) for f in files):

tests/test_cli_hooks.py

@@ -353,3 +353,18 @@ def test_uninstall_all_removes_leftover(self, tmp_path):
 
         assert result.exit_code == 0
         assert not (repo / ".git" / "hooks" / "post-commit").exists()
+
+    def test_status_survives_unreadable_leftover(self, tmp_path):
+        """codex review, #167 follow-up: an unreadable leftover must not
+        crash status (it is informational output)."""
+        repo = self._make_repo(tmp_path)
+        self._write_leftover(repo)
+
+        from unittest.mock import patch as mock_patch
+
+        with mock_patch(
+            "pathlib.Path.read_text", side_effect=PermissionError("denied")
+        ):
+            result = self._run_cli(repo, ["status"])
+
+        assert result.exit_code == 0

tests/test_init_wizard_exclude.py

@@ -51,3 +51,16 @@ def test_explain_exclude_lists_test_patterns() -> None:
 
     assert "**/*.spec.ts" in text
     assert "**/__tests__/**" in text
+
+
+def test_venv_spec_files_do_not_trigger_suggestion(tmp_path: Path) -> None:
+    """A spec file inside a venv is not the user's test — pruned like
+    detect_languages prunes venvs (codex review, GH #165 follow-up)."""
+    (tmp_path / "src").mkdir()
+    (tmp_path / "src" / "main.py").write_text("x = 1\n")
+    (tmp_path / ".venv").mkdir()
+    (tmp_path / ".venv" / "some-lib.spec.ts").write_text("describe('dep');\n")
+
+    excludes = infer_exclude_patterns(tmp_path)
+
+    assert not any(p in excludes for p in JS_TS_TEST_PATTERNS)