feat(sandbox): add built-in grep and glob tools (#1784)

* feat(sandbox): add grep and glob tools

* refactor(aio-sandbox): use native file search APIs

* fix(sandbox): address review issues in grep/glob tools

- aio_sandbox: use should_ignore_path() instead of should_ignore_name()
  for include_dirs=True branch to filter nested ignored paths correctly
- aio_sandbox: add early exit when max_results reached in glob loop
- aio_sandbox: guard entry.path.startswith(path) before stripping prefix
- aio_sandbox: validate regex locally before sending to remote API
- search: skip lines exceeding max_line_chars to prevent ReDoS
- search: remove resolve() syscall in os.walk loop
- tools: avoid double get_thread_data() call in glob_tool/grep_tool
- tests: add 6 new cases covering the above code paths
- tests: patch get_app_config in truncation test to isolate config

* Fix sandbox grep/glob review feedback

* Remove unrelated Langfuse RFC from PR

.gitignore

@@ -54,4 +54,5 @@ web/
 # Deployment artifacts
 backend/Dockerfile.langgraph
 config.yaml.bak
-.playwright-mcp
+.playwright-mcp
+.gstack/

backend/docs/rfc-grep-glob-tools.md

@@ -0,0 +1,446 @@
+# [RFC] 在 DeerFlow 中增加 `grep` 与 `glob` 文件搜索工具
+
+## Summary
+
+我认为这个方向是对的，而且值得做。
+
+如果 DeerFlow 想更接近 Claude Code 这类 coding agent 的实际工作流，仅有 `ls` / `read_file` / `write_file` / `str_replace` 还不够。模型在进入修改前，通常还需要两类能力：
+
+- `glob`: 快速按路径模式找文件
+- `grep`: 快速按内容模式找候选位置
+
+这两类工具的价值，不是“功能上 bash 也能做”，而是它们能以更低 token 成本、更强约束、更稳定的输出格式，替代模型频繁走 `bash find` / `bash grep` / `rg` 的习惯。
+
+但前提是实现方式要对：**它们应该是只读、结构化、受限、可审计的原生工具，而不是对 shell 命令的简单包装。**
+
+## Problem
+
+当前 DeerFlow 的文件工具层主要覆盖：
+
+- `ls`: 浏览目录结构
+- `read_file`: 读取文件内容
+- `write_file`: 写文件
+- `str_replace`: 做局部字符串替换
+- `bash`: 兜底执行命令
+
+这套能力能完成任务，但在代码库探索阶段效率不高。
+
+典型问题：
+
+1. 模型想找 “所有 `*.tsx` 的 page 文件” 时，只能反复 `ls` 多层目录，或者退回 `bash find`
+2. 模型想找 “某个 symbol / 文案 / 配置键在哪里出现” 时，只能逐文件 `read_file`，或者退回 `bash grep` / `rg`
+3. 一旦退回 `bash`，工具调用就失去结构化输出，结果也更难做裁剪、分页、审计和跨 sandbox 一致化
+4. 对没有开启 host bash 的本地模式，`bash` 甚至可能不可用，此时缺少足够强的只读检索能力
+
+结论：DeerFlow 现在缺的不是“再多一个 shell 命令”，而是**文件系统检索层**。
+
+## Goals
+
+- 为 agent 提供稳定的路径搜索和内容搜索能力
+- 减少对 `bash` 的依赖，特别是在仓库探索阶段
+- 保持与现有 sandbox 安全模型一致
+- 输出格式结构化，便于模型后续串联 `read_file` / `str_replace`
+- 让本地 sandbox、容器 sandbox、未来 MCP 文件系统工具都能遵守同一语义
+
+## Non-Goals
+
+- 不做通用 shell 兼容层
+- 不暴露完整 grep/find/rg CLI 语法
+- 不在第一版支持二进制检索、复杂 PCRE 特性、上下文窗口高亮渲染等重功能
+- 不把它做成“任意磁盘搜索”，仍然只允许在 DeerFlow 已授权的路径内执行
+
+## Why This Is Worth Doing
+
+参考 Claude Code 这一类 agent 的设计思路，`glob` 和 `grep` 的核心价值不是新能力本身，而是把“探索代码库”的常见动作从开放式 shell 降到受控工具层。
+
+这样有几个直接收益：
+
+1. **更低的模型负担**
+   模型不需要自己拼 `find`, `grep`, `rg`, `xargs`, quoting 等命令细节。
+
+2. **更稳定的跨环境行为**
+   本地、Docker、AIO sandbox 不必依赖容器里是否装了 `rg`，也不会因为 shell 差异导致行为漂移。
+
+3. **更强的安全与审计**
+   调用参数就是“搜索什么、在哪搜、最多返回多少”，天然比任意命令更容易审计和限流。
+
+4. **更好的 token 效率**
+   `grep` 返回的是命中摘要而不是整段文件，模型只对少数候选路径再调用 `read_file`。
+
+5. **对 `tool_search` 友好**
+   当 DeerFlow 持续扩展工具集时，`grep` / `glob` 会成为非常高频的基础工具，值得保留为 built-in，而不是让模型总是退回通用 bash。
+
+## Proposal
+
+增加两个 built-in sandbox tools：
+
+- `glob`
+- `grep`
+
+推荐继续放在：
+
+- `backend/packages/harness/deerflow/sandbox/tools.py`
+
+并在 `config.example.yaml` 中默认加入 `file:read` 组。
+
+### 1. `glob` 工具
+
+用途：按路径模式查找文件或目录。
+
+建议 schema：
+
+```python
+@tool("glob", parse_docstring=True)
+def glob_tool(
+    runtime: ToolRuntime[ContextT, ThreadState],
+    description: str,
+    pattern: str,
+    path: str,
+    include_dirs: bool = False,
+    max_results: int = 200,
+) -> str:
+    ...
+```
+
+参数语义：
+
+- `description`: 与现有工具保持一致
+- `pattern`: glob 模式，例如 `**/*.py`、`src/**/test_*.ts`
+- `path`: 搜索根目录，必须是绝对路径
+- `include_dirs`: 是否返回目录
+- `max_results`: 最大返回条数，防止一次性打爆上下文
+
+建议返回格式：
+
+```text
+Found 3 paths under /mnt/user-data/workspace
+1. /mnt/user-data/workspace/backend/app.py
+2. /mnt/user-data/workspace/backend/tests/test_app.py
+3. /mnt/user-data/workspace/scripts/build.py
+```
+
+如果后续想更适合前端消费，也可以改成 JSON 字符串；但第一版为了兼容现有工具风格，返回可读文本即可。
+
+### 2. `grep` 工具
+
+用途：按内容模式搜索文件，返回命中位置摘要。
+
+建议 schema：
+
+```python
+@tool("grep", parse_docstring=True)
+def grep_tool(
+    runtime: ToolRuntime[ContextT, ThreadState],
+    description: str,
+    pattern: str,
+    path: str,
+    glob: str | None = None,
+    literal: bool = False,
+    case_sensitive: bool = False,
+    max_results: int = 100,
+) -> str:
+    ...
+```
+
+参数语义：
+
+- `pattern`: 搜索词或正则
+- `path`: 搜索根目录，必须是绝对路径
+- `glob`: 可选路径过滤，例如 `**/*.py`
+- `literal`: 为 `True` 时按普通字符串匹配，不解释为正则
+- `case_sensitive`: 是否大小写敏感
+- `max_results`: 最大返回命中数，不是文件数
+
+建议返回格式：
+
+```text
+Found 4 matches under /mnt/user-data/workspace
+/mnt/user-data/workspace/backend/config.py:12: TOOL_GROUPS = [...]
+/mnt/user-data/workspace/backend/config.py:48: def load_tool_config(...):
+/mnt/user-data/workspace/backend/tools.py:91: "tool_groups"
+/mnt/user-data/workspace/backend/tests/test_config.py:22: assert "tool_groups" in data
+```
+
+第一版建议只返回：
+
+- 文件路径
+- 行号
+- 命中行摘要
+
+不返回上下文块，避免结果过大。模型如果需要上下文，再调用 `read_file(path, start_line, end_line)`。
+
+## Design Principles
+
+### A. 不做 shell wrapper
+
+不建议把 `grep` 实现为：
+
+```python
+subprocess.run("grep ...")
+```
+
+也不建议在容器里直接拼 `find` / `rg` 命令。
+
+原因：
+
+- 会引入 shell quoting 和注入面
+- 会依赖不同 sandbox 内镜像是否安装同一套命令
+- Windows / macOS / Linux 行为不一致
+- 很难稳定控制输出条数与格式
+
+正确方向是：
+
+- `glob` 使用 Python 标准库路径遍历
+- `grep` 使用 Python 逐文件扫描
+- 输出由 DeerFlow 自己格式化
+
+如果未来为了性能考虑要优先调用 `rg`，也应该封装在 provider 内部，并保证外部语义不变，而不是把 CLI 暴露给模型。
+
+### B. 继续沿用 DeerFlow 的路径权限模型
+
+这两个工具必须复用当前 `ls` / `read_file` 的路径校验逻辑：
+
+- 本地模式走 `validate_local_tool_path(..., read_only=True)`
+- 支持 `/mnt/skills/...`
+- 支持 `/mnt/acp-workspace/...`
+- 支持 thread workspace / uploads / outputs 的虚拟路径解析
+- 明确拒绝越权路径与 path traversal
+
+也就是说，它们属于 **file:read**，不是 `bash` 的替代越权入口。
+
+### C. 结果必须硬限制
+
+没有硬限制的 `glob` / `grep` 很容易炸上下文。
+
+建议第一版至少限制：
+
+- `glob.max_results` 默认 200，最大 1000
+- `grep.max_results` 默认 100，最大 500
+- 单行摘要最大长度，例如 200 字符
+- 二进制文件跳过
+- 超大文件跳过，例如单文件大于 1 MB 或按配置控制
+
+此外，命中数超过阈值时应返回：
+
+- 已展示的条数
+- 被截断的事实
+- 建议用户缩小搜索范围
+
+例如：
+
+```text
+Found more than 100 matches, showing first 100. Narrow the path or add a glob filter.
+```
+
+### D. 工具语义要彼此互补
+
+推荐模型工作流应该是：
+
+1. `glob` 找候选文件
+2. `grep` 找候选位置
+3. `read_file` 读局部上下文
+4. `str_replace` / `write_file` 执行修改
+
+这样工具边界清晰，也更利于 prompt 中教模型形成稳定习惯。
+
+## Implementation Approach
+
+## Option A: 直接在 `sandbox/tools.py` 实现第一版
+
+这是我推荐的起步方案。
+
+做法：
+
+- 在 `sandbox/tools.py` 新增 `glob_tool` 与 `grep_tool`
+- 在 local sandbox 场景直接使用 Python 文件系统 API
+- 在非 local sandbox 场景，优先也通过 DeerFlow 自己控制的路径访问层实现
+
+优点：
+
+- 改动小
+- 能尽快验证 agent 效果
+- 不需要先改 `Sandbox` 抽象
+
+缺点：
+
+- `tools.py` 会继续变胖
+- 如果未来想在 provider 侧做性能优化，需要再抽象一次
+
+## Option B: 先扩展 `Sandbox` 抽象
+
+例如新增：
+
+```python
+class Sandbox(ABC):
+    def glob(self, path: str, pattern: str, include_dirs: bool = False, max_results: int = 200) -> list[str]:
+        ...
+
+    def grep(
+        self,
+        path: str,
+        pattern: str,
+        *,
+        glob: str | None = None,
+        literal: bool = False,
+        case_sensitive: bool = False,
+        max_results: int = 100,
+    ) -> list[GrepMatch]:
+        ...
+```
+
+优点：
+
+- 抽象更干净
+- 容器 / 远程 sandbox 可以各自优化
+
+缺点：
+
+- 首次引入成本更高
+- 需要同步改所有 sandbox provider
+
+结论：
+
+**第一版建议走 Option A，等工具价值验证后再下沉到 `Sandbox` 抽象层。**
+
+## Detailed Behavior
+
+### `glob` 行为
+
+- 输入根目录不存在：返回清晰错误
+- 根路径不是目录：返回清晰错误
+- 模式非法：返回清晰错误
+- 结果为空：返回 `No files matched`
+- 默认忽略项应尽量与当前 `list_dir` 对齐，例如：
+  - `.git`
+  - `node_modules`
+  - `__pycache__`
+  - `.venv`
+  - 构建产物目录
+
+这里建议抽一个共享 ignore 集，避免 `ls` 与 `glob` 结果风格不一致。
+
+### `grep` 行为
+
+- 默认只扫描文本文件
+- 检测到二进制文件直接跳过
+- 对超大文件直接跳过或只扫前 N KB
+- regex 编译失败时返回参数错误
+- 输出中的路径继续使用虚拟路径，而不是暴露宿主真实路径
+- 建议默认按文件路径、行号排序，保持稳定输出
+
+## Prompting Guidance
+
+如果引入这两个工具，建议同步更新系统提示中的文件操作建议：
+
+- 查找文件名模式时优先用 `glob`
+- 查找代码符号、配置项、文案时优先用 `grep`
+- 只有在工具不足以完成目标时才退回 `bash`
+
+否则模型仍会习惯性先调用 `bash`。
+
+## Risks
+
+### 1. 与 `bash` 能力重叠
+
+这是事实，但不是问题。
+
+`ls` 和 `read_file` 也都能被 `bash` 替代，但我们仍然保留它们，因为结构化工具更适合 agent。
+
+### 2. 性能问题
+
+在大仓库上，纯 Python `grep` 可能比 `rg` 慢。
+
+缓解方式：
+
+- 第一版先加结果上限和文件大小上限
+- 路径上强制要求 root path
+- 提供 `glob` 过滤缩小扫描范围
+- 后续如有必要，在 provider 内部做 `rg` 优化，但保持同一 schema
+
+### 3. 忽略规则不一致
+
+如果 `ls` 能看到的路径，`glob` 却看不到，模型会困惑。
+
+缓解方式：
+
+- 统一 ignore 规则
+- 在文档里明确“默认跳过常见依赖和构建目录”
+
+### 4. 正则搜索过于复杂
+
+如果第一版就支持大量 grep 方言，边界会很乱。
+
+缓解方式：
+
+- 第一版只支持 Python `re`
+- 并提供 `literal=True` 的简单模式
+
+## Alternatives Considered
+
+### A. 不增加工具，完全依赖 `bash`
+
+不推荐。
+
+这会让 DeerFlow 在代码探索体验上持续落后，也削弱无 bash 或受限 bash 场景下的能力。
+
+### B. 只加 `glob`，不加 `grep`
+
+不推荐。
+
+只解决“找文件”，没有解决“找位置”。模型最终还是会退回 `bash grep`。
+
+### C. 只加 `grep`，不加 `glob`
+
+也不推荐。
+
+`grep` 缺少路径模式过滤时，扫描范围经常太大；`glob` 是它的天然前置工具。
+
+### D. 直接接入 MCP filesystem server 的搜索能力
+
+短期不推荐作为主路径。
+
+MCP 可以是补充，但 `glob` / `grep` 作为 DeerFlow 的基础 coding tool，最好仍然是 built-in，这样才能在默认安装中稳定可用。
+
+## Acceptance Criteria
+
+- `config.example.yaml` 中可默认启用 `glob` 与 `grep`
+- 两个工具归属 `file:read` 组
+- 本地 sandbox 下严格遵守现有路径权限
+- 输出不泄露宿主机真实路径
+- 大结果集会被截断并明确提示
+- 模型可以通过 `glob -> grep -> read_file -> str_replace` 完成典型改码流
+- 在禁用 host bash 的本地模式下，仓库探索能力明显提升
+
+## Rollout Plan
+
+1. 在 `sandbox/tools.py` 中实现 `glob_tool` 与 `grep_tool`
+2. 抽取与 `list_dir` 一致的 ignore 规则，避免行为漂移
+3. 在 `config.example.yaml` 默认加入工具配置
+4. 为本地路径校验、虚拟路径映射、结果截断、二进制跳过补测试
+5. 更新 README / backend docs / prompt guidance
+6. 收集实际 agent 调用数据，再决定是否下沉到 `Sandbox` 抽象
+
+## Suggested Config
+
+```yaml
+tools:
+  - name: glob
+    group: file:read
+    use: deerflow.sandbox.tools:glob_tool
+
+  - name: grep
+    group: file:read
+    use: deerflow.sandbox.tools:grep_tool
+```
+
+## Final Recommendation
+
+结论是：**可以加，而且应该加。**
+
+但我会明确卡三个边界：
+
+1. `grep` / `glob` 必须是 built-in 的只读结构化工具
+2. 第一版不要做 shell wrapper，不要把 CLI 方言直接暴露给模型
+3. 先在 `sandbox/tools.py` 验证价值，再考虑是否下沉到 `Sandbox` provider 抽象
+
+如果按这个方向做，它会明显提升 DeerFlow 在 coding / repo exploration 场景下的可用性，而且风险可控。

backend/packages/harness/deerflow/community/aio_sandbox/aio_sandbox.py

@@ -7,6 +7,7 @@ import uuid
 from agent_sandbox import Sandbox as AioSandboxClient
 
 from deerflow.sandbox.sandbox import Sandbox
+from deerflow.sandbox.search import GrepMatch, path_matches, should_ignore_path, truncate_line
 
 logger = logging.getLogger(__name__)
 
@@ -135,6 +136,86 @@ class AioSandbox(Sandbox):
                 logger.error(f"Failed to write file in sandbox: {e}")
                 raise
 
+    def glob(self, path: str, pattern: str, *, include_dirs: bool = False, max_results: int = 200) -> tuple[list[str], bool]:
+        if not include_dirs:
+            result = self._client.file.find_files(path=path, glob=pattern)
+            files = result.data.files if result.data and result.data.files else []
+            filtered = [file_path for file_path in files if not should_ignore_path(file_path)]
+            truncated = len(filtered) > max_results
+            return filtered[:max_results], truncated
+
+        result = self._client.file.list_path(path=path, recursive=True, show_hidden=False)
+        entries = result.data.files if result.data and result.data.files else []
+        matches: list[str] = []
+        root_path = path.rstrip("/") or "/"
+        root_prefix = root_path if root_path == "/" else f"{root_path}/"
+        for entry in entries:
+            if entry.path != root_path and not entry.path.startswith(root_prefix):
+                continue
+            if should_ignore_path(entry.path):
+                continue
+            rel_path = entry.path[len(root_path) :].lstrip("/")
+            if path_matches(pattern, rel_path):
+                matches.append(entry.path)
+                if len(matches) >= max_results:
+                    return matches, True
+        return matches, False
+
+    def grep(
+        self,
+        path: str,
+        pattern: str,
+        *,
+        glob: str | None = None,
+        literal: bool = False,
+        case_sensitive: bool = False,
+        max_results: int = 100,
+    ) -> tuple[list[GrepMatch], bool]:
+        import re as _re
+
+        regex_source = _re.escape(pattern) if literal else pattern
+        # Validate the pattern locally so an invalid regex raises re.error
+        # (caught by grep_tool's except re.error handler) rather than a
+        # generic remote API error.
+        _re.compile(regex_source, 0 if case_sensitive else _re.IGNORECASE)
+        regex = regex_source if case_sensitive else f"(?i){regex_source}"
+
+        if glob is not None:
+            find_result = self._client.file.find_files(path=path, glob=glob)
+            candidate_paths = find_result.data.files if find_result.data and find_result.data.files else []
+        else:
+            list_result = self._client.file.list_path(path=path, recursive=True, show_hidden=False)
+            entries = list_result.data.files if list_result.data and list_result.data.files else []
+            candidate_paths = [entry.path for entry in entries if not entry.is_directory]
+
+        matches: list[GrepMatch] = []
+        truncated = False
+
+        for file_path in candidate_paths:
+            if should_ignore_path(file_path):
+                continue
+
+            search_result = self._client.file.search_in_file(file=file_path, regex=regex)
+            data = search_result.data
+            if data is None:
+                continue
+
+            line_numbers = data.line_numbers or []
+            matched_lines = data.matches or []
+            for line_number, line in zip(line_numbers, matched_lines):
+                matches.append(
+                    GrepMatch(
+                        path=file_path,
+                        line_number=line_number if isinstance(line_number, int) else 0,
+                        line=truncate_line(line),
+                    )
+                )
+                if len(matches) >= max_results:
+                    truncated = True
+                    return matches, truncated
+
+        return matches, truncated
+
     def update_file(self, path: str, content: bytes) -> None:
         """Update a file with binary content in the sandbox.
 

backend/packages/harness/deerflow/sandbox/local/list_dir.py

@@ -1,72 +1,6 @@
-import fnmatch
 from pathlib import Path
 
-IGNORE_PATTERNS = [
+from deerflow.sandbox.search import should_ignore_name
-    # Version Control
-    ".git",
-    ".svn",
-    ".hg",
-    ".bzr",
-    # Dependencies
-    "node_modules",
-    "__pycache__",
-    ".venv",
-    "venv",
-    ".env",
-    "env",
-    ".tox",
-    ".nox",
-    ".eggs",
-    "*.egg-info",
-    "site-packages",
-    # Build outputs
-    "dist",
-    "build",
-    ".next",
-    ".nuxt",
-    ".output",
-    ".turbo",
-    "target",
-    "out",
-    # IDE & Editor
-    ".idea",
-    ".vscode",
-    "*.swp",
-    "*.swo",
-    "*~",
-    ".project",
-    ".classpath",
-    ".settings",
-    # OS generated
-    ".DS_Store",
-    "Thumbs.db",
-    "desktop.ini",
-    "*.lnk",
-    # Logs & temp files
-    "*.log",
-    "*.tmp",
-    "*.temp",
-    "*.bak",
-    "*.cache",
-    ".cache",
-    "logs",
-    # Coverage & test artifacts
-    ".coverage",
-    "coverage",
-    ".nyc_output",
-    "htmlcov",
-    ".pytest_cache",
-    ".mypy_cache",
-    ".ruff_cache",
-]
-
-
-def _should_ignore(name: str) -> bool:
-    """Check if a file/directory name matches any ignore pattern."""
-    for pattern in IGNORE_PATTERNS:
-        if fnmatch.fnmatch(name, pattern):
-            return True
-    return False
 
 
 def list_dir(path: str, max_depth: int = 2) -> list[str]:
@@ -95,7 +29,7 @@ def list_dir(path: str, max_depth: int = 2) -> list[str]:
 
         try:
             for item in current_path.iterdir():
-                if _should_ignore(item.name):
+                if should_ignore_name(item.name):
                     continue
 
                 post_fix = "/" if item.is_dir() else ""

backend/packages/harness/deerflow/sandbox/local/local_sandbox.py

@@ -6,6 +6,7 @@ from pathlib import Path
 
 from deerflow.sandbox.local.list_dir import list_dir
 from deerflow.sandbox.sandbox import Sandbox
+from deerflow.sandbox.search import GrepMatch, find_glob_matches, find_grep_matches
 
 
 class LocalSandbox(Sandbox):
@@ -259,6 +260,39 @@ class LocalSandbox(Sandbox):
             # Re-raise with the original path for clearer error messages, hiding internal resolved paths
             raise type(e)(e.errno, e.strerror, path) from None
 
+    def glob(self, path: str, pattern: str, *, include_dirs: bool = False, max_results: int = 200) -> tuple[list[str], bool]:
+        resolved_path = Path(self._resolve_path(path))
+        matches, truncated = find_glob_matches(resolved_path, pattern, include_dirs=include_dirs, max_results=max_results)
+        return [self._reverse_resolve_path(match) for match in matches], truncated
+
+    def grep(
+        self,
+        path: str,
+        pattern: str,
+        *,
+        glob: str | None = None,
+        literal: bool = False,
+        case_sensitive: bool = False,
+        max_results: int = 100,
+    ) -> tuple[list[GrepMatch], bool]:
+        resolved_path = Path(self._resolve_path(path))
+        matches, truncated = find_grep_matches(
+            resolved_path,
+            pattern,
+            glob_pattern=glob,
+            literal=literal,
+            case_sensitive=case_sensitive,
+            max_results=max_results,
+        )
+        return [
+            GrepMatch(
+                path=self._reverse_resolve_path(match.path),
+                line_number=match.line_number,
+                line=match.line,
+            )
+            for match in matches
+        ], truncated
+
     def update_file(self, path: str, content: bytes) -> None:
         resolved_path = self._resolve_path(path)
         try:

backend/packages/harness/deerflow/sandbox/sandbox.py

@@ -1,5 +1,7 @@
 from abc import ABC, abstractmethod
 
+from deerflow.sandbox.search import GrepMatch
+
 
 class Sandbox(ABC):
     """Abstract base class for sandbox environments"""
@@ -61,6 +63,25 @@ class Sandbox(ABC):
         """
         pass
 
+    @abstractmethod
+    def glob(self, path: str, pattern: str, *, include_dirs: bool = False, max_results: int = 200) -> tuple[list[str], bool]:
+        """Find paths that match a glob pattern under a root directory."""
+        pass
+
+    @abstractmethod
+    def grep(
+        self,
+        path: str,
+        pattern: str,
+        *,
+        glob: str | None = None,
+        literal: bool = False,
+        case_sensitive: bool = False,
+        max_results: int = 100,
+    ) -> tuple[list[GrepMatch], bool]:
+        """Search for matches inside text files under a directory."""
+        pass
+
     @abstractmethod
     def update_file(self, path: str, content: bytes) -> None:
         """Update a file with binary content.

backend/packages/harness/deerflow/sandbox/search.py

@@ -0,0 +1,210 @@
+import fnmatch
+import os
+import re
+from dataclasses import dataclass
+from pathlib import Path, PurePosixPath
+
+IGNORE_PATTERNS = [
+    ".git",
+    ".svn",
+    ".hg",
+    ".bzr",
+    "node_modules",
+    "__pycache__",
+    ".venv",
+    "venv",
+    ".env",
+    "env",
+    ".tox",
+    ".nox",
+    ".eggs",
+    "*.egg-info",
+    "site-packages",
+    "dist",
+    "build",
+    ".next",
+    ".nuxt",
+    ".output",
+    ".turbo",
+    "target",
+    "out",
+    ".idea",
+    ".vscode",
+    "*.swp",
+    "*.swo",
+    "*~",
+    ".project",
+    ".classpath",
+    ".settings",
+    ".DS_Store",
+    "Thumbs.db",
+    "desktop.ini",
+    "*.lnk",
+    "*.log",
+    "*.tmp",
+    "*.temp",
+    "*.bak",
+    "*.cache",
+    ".cache",
+    "logs",
+    ".coverage",
+    "coverage",
+    ".nyc_output",
+    "htmlcov",
+    ".pytest_cache",
+    ".mypy_cache",
+    ".ruff_cache",
+]
+
+DEFAULT_MAX_FILE_SIZE_BYTES = 1_000_000
+DEFAULT_LINE_SUMMARY_LENGTH = 200
+
+
+@dataclass(frozen=True)
+class GrepMatch:
+    path: str
+    line_number: int
+    line: str
+
+
+def should_ignore_name(name: str) -> bool:
+    for pattern in IGNORE_PATTERNS:
+        if fnmatch.fnmatch(name, pattern):
+            return True
+    return False
+
+
+def should_ignore_path(path: str) -> bool:
+    return any(should_ignore_name(segment) for segment in path.replace("\\", "/").split("/") if segment)
+
+
+def path_matches(pattern: str, rel_path: str) -> bool:
+    path = PurePosixPath(rel_path)
+    if path.match(pattern):
+        return True
+    if pattern.startswith("**/"):
+        return path.match(pattern[3:])
+    return False
+
+
+def truncate_line(line: str, max_chars: int = DEFAULT_LINE_SUMMARY_LENGTH) -> str:
+    line = line.rstrip("\n\r")
+    if len(line) <= max_chars:
+        return line
+    return line[: max_chars - 3] + "..."
+
+
+def is_binary_file(path: Path, sample_size: int = 8192) -> bool:
+    try:
+        with path.open("rb") as handle:
+            return b"\0" in handle.read(sample_size)
+    except OSError:
+        return True
+
+
+def find_glob_matches(root: Path, pattern: str, *, include_dirs: bool = False, max_results: int = 200) -> tuple[list[str], bool]:
+    matches: list[str] = []
+    truncated = False
+    root = root.resolve()
+
+    if not root.exists():
+        raise FileNotFoundError(root)
+    if not root.is_dir():
+        raise NotADirectoryError(root)
+
+    for current_root, dirs, files in os.walk(root):
+        dirs[:] = [name for name in dirs if not should_ignore_name(name)]
+        # root is already resolved; os.walk builds current_root by joining under root,
+        # so relative_to() works without an extra stat()/resolve() per directory.
+        rel_dir = Path(current_root).relative_to(root)
+
+        if include_dirs:
+            for name in dirs:
+                rel_path = (rel_dir / name).as_posix()
+                if path_matches(pattern, rel_path):
+                    matches.append(str(Path(current_root) / name))
+                    if len(matches) >= max_results:
+                        truncated = True
+                        return matches, truncated
+
+        for name in files:
+            if should_ignore_name(name):
+                continue
+            rel_path = (rel_dir / name).as_posix()
+            if path_matches(pattern, rel_path):
+                matches.append(str(Path(current_root) / name))
+                if len(matches) >= max_results:
+                    truncated = True
+                    return matches, truncated
+
+    return matches, truncated
+
+
+def find_grep_matches(
+    root: Path,
+    pattern: str,
+    *,
+    glob_pattern: str | None = None,
+    literal: bool = False,
+    case_sensitive: bool = False,
+    max_results: int = 100,
+    max_file_size: int = DEFAULT_MAX_FILE_SIZE_BYTES,
+    line_summary_length: int = DEFAULT_LINE_SUMMARY_LENGTH,
+) -> tuple[list[GrepMatch], bool]:
+    matches: list[GrepMatch] = []
+    truncated = False
+    root = root.resolve()
+
+    if not root.exists():
+        raise FileNotFoundError(root)
+    if not root.is_dir():
+        raise NotADirectoryError(root)
+
+    regex_source = re.escape(pattern) if literal else pattern
+    flags = 0 if case_sensitive else re.IGNORECASE
+    regex = re.compile(regex_source, flags)
+
+    # Skip lines longer than this to prevent ReDoS on minified / no-newline files.
+    _max_line_chars = line_summary_length * 10
+
+    for current_root, dirs, files in os.walk(root):
+        dirs[:] = [name for name in dirs if not should_ignore_name(name)]
+        rel_dir = Path(current_root).relative_to(root)
+
+        for name in files:
+            if should_ignore_name(name):
+                continue
+
+            candidate_path = Path(current_root) / name
+            rel_path = (rel_dir / name).as_posix()
+
+            if glob_pattern is not None and not path_matches(glob_pattern, rel_path):
+                continue
+
+            try:
+                if candidate_path.is_symlink():
+                    continue
+                file_path = candidate_path.resolve()
+                if not file_path.is_relative_to(root):
+                    continue
+                if file_path.stat().st_size > max_file_size or is_binary_file(file_path):
+                    continue
+                with file_path.open(encoding="utf-8", errors="replace") as handle:
+                    for line_number, line in enumerate(handle, start=1):
+                        if len(line) > _max_line_chars:
+                            continue
+                        if regex.search(line):
+                            matches.append(
+                                GrepMatch(
+                                    path=str(file_path),
+                                    line_number=line_number,
+                                    line=truncate_line(line, line_summary_length),
+                                )
+                            )
+                            if len(matches) >= max_results:
+                                truncated = True
+                                return matches, truncated
+            except OSError:
+                continue
+
+    return matches, truncated

backend/packages/harness/deerflow/sandbox/tools.py

@@ -7,6 +7,7 @@ from langchain.tools import ToolRuntime, tool
 from langgraph.typing import ContextT
 
 from deerflow.agents.thread_state import ThreadDataState, ThreadState
+from deerflow.config import get_app_config
 from deerflow.config.paths import VIRTUAL_PATH_PREFIX
 from deerflow.sandbox.exceptions import (
     SandboxError,
@@ -16,6 +17,7 @@ from deerflow.sandbox.exceptions import (
 from deerflow.sandbox.file_operation_lock import get_file_operation_lock
 from deerflow.sandbox.sandbox import Sandbox
 from deerflow.sandbox.sandbox_provider import get_sandbox_provider
+from deerflow.sandbox.search import GrepMatch
 from deerflow.sandbox.security import LOCAL_HOST_BASH_DISABLED_MESSAGE, is_host_bash_allowed
 
 _ABSOLUTE_PATH_PATTERN = re.compile(r"(?<![:\w])(?<!:/)/(?:[^\s\"'`;&|<>()]+)")
@@ -31,6 +33,10 @@ _LOCAL_BASH_SYSTEM_PATH_PREFIXES = (
 
 _DEFAULT_SKILLS_CONTAINER_PATH = "/mnt/skills"
 _ACP_WORKSPACE_VIRTUAL_PATH = "/mnt/acp-workspace"
+_DEFAULT_GLOB_MAX_RESULTS = 200
+_MAX_GLOB_MAX_RESULTS = 1000
+_DEFAULT_GREP_MAX_RESULTS = 100
+_MAX_GREP_MAX_RESULTS = 500
 
 
 def _get_skills_container_path() -> str:
@@ -245,6 +251,69 @@ def _get_mcp_allowed_paths() -> list[str]:
     return allowed_paths
 
 
+def _get_tool_config_int(name: str, key: str, default: int) -> int:
+    try:
+        tool_config = get_app_config().get_tool_config(name)
+        if tool_config is not None and key in tool_config.model_extra:
+            value = tool_config.model_extra.get(key)
+            if isinstance(value, int):
+                return value
+    except Exception:
+        pass
+    return default
+
+
+def _clamp_max_results(value: int, *, default: int, upper_bound: int) -> int:
+    if value <= 0:
+        return default
+    return min(value, upper_bound)
+
+
+def _resolve_max_results(name: str, requested: int, *, default: int, upper_bound: int) -> int:
+    requested_max_results = _clamp_max_results(requested, default=default, upper_bound=upper_bound)
+    configured_max_results = _clamp_max_results(
+        _get_tool_config_int(name, "max_results", default),
+        default=default,
+        upper_bound=upper_bound,
+    )
+    return min(requested_max_results, configured_max_results)
+
+
+def _resolve_local_read_path(path: str, thread_data: ThreadDataState) -> str:
+    validate_local_tool_path(path, thread_data, read_only=True)
+    if _is_skills_path(path):
+        return _resolve_skills_path(path)
+    if _is_acp_workspace_path(path):
+        return _resolve_acp_workspace_path(path, _extract_thread_id_from_thread_data(thread_data))
+    return _resolve_and_validate_user_data_path(path, thread_data)
+
+
+def _format_glob_results(root_path: str, matches: list[str], truncated: bool) -> str:
+    if not matches:
+        return f"No files matched under {root_path}"
+
+    lines = [f"Found {len(matches)} paths under {root_path}"]
+    if truncated:
+        lines[0] += f" (showing first {len(matches)})"
+    lines.extend(f"{index}. {path}" for index, path in enumerate(matches, start=1))
+    if truncated:
+        lines.append("Results truncated. Narrow the path or pattern to see fewer matches.")
+    return "\n".join(lines)
+
+
+def _format_grep_results(root_path: str, matches: list[GrepMatch], truncated: bool) -> str:
+    if not matches:
+        return f"No matches found under {root_path}"
+
+    lines = [f"Found {len(matches)} matches under {root_path}"]
+    if truncated:
+        lines[0] += f" (showing first {len(matches)})"
+    lines.extend(f"{match.path}:{match.line_number}: {match.line}" for match in matches)
+    if truncated:
+        lines.append("Results truncated. Narrow the path or add a glob filter.")
+    return "\n".join(lines)
+
+
 def _path_variants(path: str) -> set[str]:
     return {path, path.replace("\\", "/"), path.replace("/", "\\")}
 
@@ -901,6 +970,126 @@ def ls_tool(runtime: ToolRuntime[ContextT, ThreadState], description: str, path:
         return f"Error: Unexpected error listing directory: {_sanitize_error(e, runtime)}"
 
 
+@tool("glob", parse_docstring=True)
+def glob_tool(
+    runtime: ToolRuntime[ContextT, ThreadState],
+    description: str,
+    pattern: str,
+    path: str,
+    include_dirs: bool = False,
+    max_results: int = _DEFAULT_GLOB_MAX_RESULTS,
+) -> str:
+    """Find files or directories that match a glob pattern under a root directory.
+
+    Args:
+        description: Explain why you are searching for these paths in short words. ALWAYS PROVIDE THIS PARAMETER FIRST.
+        pattern: The glob pattern to match relative to the root path, for example `**/*.py`.
+        path: The **absolute** root directory to search under.
+        include_dirs: Whether matching directories should also be returned. Default is False.
+        max_results: Maximum number of paths to return. Default is 200.
+    """
+    try:
+        sandbox = ensure_sandbox_initialized(runtime)
+        ensure_thread_directories_exist(runtime)
+        requested_path = path
+        effective_max_results = _resolve_max_results(
+            "glob",
+            max_results,
+            default=_DEFAULT_GLOB_MAX_RESULTS,
+            upper_bound=_MAX_GLOB_MAX_RESULTS,
+        )
+        thread_data = None
+        if is_local_sandbox(runtime):
+            thread_data = get_thread_data(runtime)
+            if thread_data is None:
+                raise SandboxRuntimeError("Thread data not available for local sandbox")
+            path = _resolve_local_read_path(path, thread_data)
+        matches, truncated = sandbox.glob(path, pattern, include_dirs=include_dirs, max_results=effective_max_results)
+        if thread_data is not None:
+            matches = [mask_local_paths_in_output(match, thread_data) for match in matches]
+        return _format_glob_results(requested_path, matches, truncated)
+    except SandboxError as e:
+        return f"Error: {e}"
+    except FileNotFoundError:
+        return f"Error: Directory not found: {requested_path}"
+    except NotADirectoryError:
+        return f"Error: Path is not a directory: {requested_path}"
+    except PermissionError:
+        return f"Error: Permission denied: {requested_path}"
+    except Exception as e:
+        return f"Error: Unexpected error searching paths: {_sanitize_error(e, runtime)}"
+
+
+@tool("grep", parse_docstring=True)
+def grep_tool(
+    runtime: ToolRuntime[ContextT, ThreadState],
+    description: str,
+    pattern: str,
+    path: str,
+    glob: str | None = None,
+    literal: bool = False,
+    case_sensitive: bool = False,
+    max_results: int = _DEFAULT_GREP_MAX_RESULTS,
+) -> str:
+    """Search for matching lines inside text files under a root directory.
+
+    Args:
+        description: Explain why you are searching file contents in short words. ALWAYS PROVIDE THIS PARAMETER FIRST.
+        pattern: The string or regex pattern to search for.
+        path: The **absolute** root directory to search under.
+        glob: Optional glob filter for candidate files, for example `**/*.py`.
+        literal: Whether to treat `pattern` as a plain string. Default is False.
+        case_sensitive: Whether matching is case-sensitive. Default is False.
+        max_results: Maximum number of matching lines to return. Default is 100.
+    """
+    try:
+        sandbox = ensure_sandbox_initialized(runtime)
+        ensure_thread_directories_exist(runtime)
+        requested_path = path
+        effective_max_results = _resolve_max_results(
+            "grep",
+            max_results,
+            default=_DEFAULT_GREP_MAX_RESULTS,
+            upper_bound=_MAX_GREP_MAX_RESULTS,
+        )
+        thread_data = None
+        if is_local_sandbox(runtime):
+            thread_data = get_thread_data(runtime)
+            if thread_data is None:
+                raise SandboxRuntimeError("Thread data not available for local sandbox")
+            path = _resolve_local_read_path(path, thread_data)
+        matches, truncated = sandbox.grep(
+            path,
+            pattern,
+            glob=glob,
+            literal=literal,
+            case_sensitive=case_sensitive,
+            max_results=effective_max_results,
+        )
+        if thread_data is not None:
+            matches = [
+                GrepMatch(
+                    path=mask_local_paths_in_output(match.path, thread_data),
+                    line_number=match.line_number,
+                    line=match.line,
+                )
+                for match in matches
+            ]
+        return _format_grep_results(requested_path, matches, truncated)
+    except SandboxError as e:
+        return f"Error: {e}"
+    except FileNotFoundError:
+        return f"Error: Directory not found: {requested_path}"
+    except NotADirectoryError:
+        return f"Error: Path is not a directory: {requested_path}"
+    except re.error as e:
+        return f"Error: Invalid regex pattern: {e}"
+    except PermissionError:
+        return f"Error: Permission denied: {requested_path}"
+    except Exception as e:
+        return f"Error: Unexpected error searching file contents: {_sanitize_error(e, runtime)}"
+
+
 @tool("read_file", parse_docstring=True)
 def read_file_tool(
     runtime: ToolRuntime[ContextT, ThreadState],

backend/tests/test_sandbox_search_tools.py

@@ -0,0 +1,393 @@
+from types import SimpleNamespace
+from unittest.mock import patch
+
+from deerflow.community.aio_sandbox.aio_sandbox import AioSandbox
+from deerflow.sandbox.local.local_sandbox import LocalSandbox
+from deerflow.sandbox.search import GrepMatch, find_glob_matches, find_grep_matches
+from deerflow.sandbox.tools import glob_tool, grep_tool
+
+
+def _make_runtime(tmp_path):
+    workspace = tmp_path / "workspace"
+    uploads = tmp_path / "uploads"
+    outputs = tmp_path / "outputs"
+    workspace.mkdir()
+    uploads.mkdir()
+    outputs.mkdir()
+    return SimpleNamespace(
+        state={
+            "sandbox": {"sandbox_id": "local"},
+            "thread_data": {
+                "workspace_path": str(workspace),
+                "uploads_path": str(uploads),
+                "outputs_path": str(outputs),
+            },
+        },
+        context={"thread_id": "thread-1"},
+    )
+
+
+def test_glob_tool_returns_virtual_paths_and_ignores_common_dirs(tmp_path, monkeypatch) -> None:
+    runtime = _make_runtime(tmp_path)
+    workspace = tmp_path / "workspace"
+    (workspace / "app.py").write_text("print('hi')\n", encoding="utf-8")
+    (workspace / "pkg").mkdir()
+    (workspace / "pkg" / "util.py").write_text("print('util')\n", encoding="utf-8")
+    (workspace / "node_modules").mkdir()
+    (workspace / "node_modules" / "skip.py").write_text("ignored\n", encoding="utf-8")
+
+    monkeypatch.setattr("deerflow.sandbox.tools.ensure_sandbox_initialized", lambda runtime: LocalSandbox(id="local"))
+
+    result = glob_tool.func(
+        runtime=runtime,
+        description="find python files",
+        pattern="**/*.py",
+        path="/mnt/user-data/workspace",
+    )
+
+    assert "/mnt/user-data/workspace/app.py" in result
+    assert "/mnt/user-data/workspace/pkg/util.py" in result
+    assert "node_modules" not in result
+    assert str(workspace) not in result
+
+
+def test_glob_tool_supports_skills_virtual_paths(tmp_path, monkeypatch) -> None:
+    runtime = _make_runtime(tmp_path)
+    skills_dir = tmp_path / "skills"
+    (skills_dir / "public" / "demo").mkdir(parents=True)
+    (skills_dir / "public" / "demo" / "SKILL.md").write_text("# Demo\n", encoding="utf-8")
+
+    monkeypatch.setattr("deerflow.sandbox.tools.ensure_sandbox_initialized", lambda runtime: LocalSandbox(id="local"))
+
+    with (
+        patch("deerflow.sandbox.tools._get_skills_container_path", return_value="/mnt/skills"),
+        patch("deerflow.sandbox.tools._get_skills_host_path", return_value=str(skills_dir)),
+    ):
+        result = glob_tool.func(
+            runtime=runtime,
+            description="find skills",
+            pattern="**/SKILL.md",
+            path="/mnt/skills",
+        )
+
+    assert "/mnt/skills/public/demo/SKILL.md" in result
+    assert str(skills_dir) not in result
+
+
+def test_grep_tool_filters_by_glob_and_skips_binary_files(tmp_path, monkeypatch) -> None:
+    runtime = _make_runtime(tmp_path)
+    workspace = tmp_path / "workspace"
+    (workspace / "main.py").write_text("TODO = 'ship it'\nprint(TODO)\n", encoding="utf-8")
+    (workspace / "notes.txt").write_text("TODO in txt should be filtered\n", encoding="utf-8")
+    (workspace / "image.bin").write_bytes(b"\0binary TODO")
+
+    monkeypatch.setattr("deerflow.sandbox.tools.ensure_sandbox_initialized", lambda runtime: LocalSandbox(id="local"))
+
+    result = grep_tool.func(
+        runtime=runtime,
+        description="find todo references",
+        pattern="TODO",
+        path="/mnt/user-data/workspace",
+        glob="**/*.py",
+    )
+
+    assert "/mnt/user-data/workspace/main.py:1: TODO = 'ship it'" in result
+    assert "notes.txt" not in result
+    assert "image.bin" not in result
+    assert str(workspace) not in result
+
+
+def test_grep_tool_truncates_results(tmp_path, monkeypatch) -> None:
+    runtime = _make_runtime(tmp_path)
+    workspace = tmp_path / "workspace"
+    (workspace / "main.py").write_text("TODO one\nTODO two\nTODO three\n", encoding="utf-8")
+
+    monkeypatch.setattr("deerflow.sandbox.tools.ensure_sandbox_initialized", lambda runtime: LocalSandbox(id="local"))
+    # Prevent config.yaml tool config from overriding the caller-supplied max_results=2.
+    monkeypatch.setattr("deerflow.sandbox.tools.get_app_config", lambda: SimpleNamespace(get_tool_config=lambda name: None))
+
+    result = grep_tool.func(
+        runtime=runtime,
+        description="limit matches",
+        pattern="TODO",
+        path="/mnt/user-data/workspace",
+        max_results=2,
+    )
+
+    assert "Found 2 matches under /mnt/user-data/workspace (showing first 2)" in result
+    assert "TODO one" in result
+    assert "TODO two" in result
+    assert "TODO three" not in result
+    assert "Results truncated." in result
+
+
+def test_glob_tool_include_dirs_filters_nested_ignored_paths(tmp_path, monkeypatch) -> None:
+    runtime = _make_runtime(tmp_path)
+    workspace = tmp_path / "workspace"
+    (workspace / "src").mkdir()
+    (workspace / "src" / "main.py").write_text("x\n", encoding="utf-8")
+    (workspace / "node_modules").mkdir()
+    (workspace / "node_modules" / "lib").mkdir()
+
+    monkeypatch.setattr("deerflow.sandbox.tools.ensure_sandbox_initialized", lambda runtime: LocalSandbox(id="local"))
+
+    result = glob_tool.func(
+        runtime=runtime,
+        description="find dirs",
+        pattern="**",
+        path="/mnt/user-data/workspace",
+        include_dirs=True,
+    )
+
+    assert "src" in result
+    assert "node_modules" not in result
+
+
+def test_grep_tool_literal_mode(tmp_path, monkeypatch) -> None:
+    runtime = _make_runtime(tmp_path)
+    workspace = tmp_path / "workspace"
+    (workspace / "file.py").write_text("price = (a+b)\nresult = a+b\n", encoding="utf-8")
+
+    monkeypatch.setattr("deerflow.sandbox.tools.ensure_sandbox_initialized", lambda runtime: LocalSandbox(id="local"))
+
+    # literal=True should treat (a+b) as a plain string, not a regex group
+    result = grep_tool.func(
+        runtime=runtime,
+        description="literal search",
+        pattern="(a+b)",
+        path="/mnt/user-data/workspace",
+        literal=True,
+    )
+
+    assert "price = (a+b)" in result
+    assert "result = a+b" not in result
+
+
+def test_grep_tool_case_sensitive(tmp_path, monkeypatch) -> None:
+    runtime = _make_runtime(tmp_path)
+    workspace = tmp_path / "workspace"
+    (workspace / "file.py").write_text("TODO: fix\ntodo: also fix\n", encoding="utf-8")
+
+    monkeypatch.setattr("deerflow.sandbox.tools.ensure_sandbox_initialized", lambda runtime: LocalSandbox(id="local"))
+
+    result = grep_tool.func(
+        runtime=runtime,
+        description="case sensitive search",
+        pattern="TODO",
+        path="/mnt/user-data/workspace",
+        case_sensitive=True,
+    )
+
+    assert "TODO: fix" in result
+    assert "todo: also fix" not in result
+
+
+def test_grep_tool_invalid_regex_returns_error(tmp_path, monkeypatch) -> None:
+    runtime = _make_runtime(tmp_path)
+
+    monkeypatch.setattr("deerflow.sandbox.tools.ensure_sandbox_initialized", lambda runtime: LocalSandbox(id="local"))
+
+    result = grep_tool.func(
+        runtime=runtime,
+        description="bad pattern",
+        pattern="[invalid",
+        path="/mnt/user-data/workspace",
+    )
+
+    assert "Invalid regex pattern" in result
+
+
+def test_aio_sandbox_glob_include_dirs_filters_nested_ignored(monkeypatch) -> None:
+    with patch("deerflow.community.aio_sandbox.aio_sandbox.AioSandboxClient"):
+        sandbox = AioSandbox(id="test-sandbox", base_url="http://localhost:8080")
+    monkeypatch.setattr(
+        sandbox._client.file,
+        "list_path",
+        lambda **kwargs: SimpleNamespace(
+            data=SimpleNamespace(
+                files=[
+                    SimpleNamespace(name="src", path="/mnt/workspace/src"),
+                    SimpleNamespace(name="node_modules", path="/mnt/workspace/node_modules"),
+                    # child of node_modules — should be filtered via should_ignore_path
+                    SimpleNamespace(name="lib", path="/mnt/workspace/node_modules/lib"),
+                ]
+            )
+        ),
+    )
+
+    matches, truncated = sandbox.glob("/mnt/workspace", "**", include_dirs=True)
+
+    assert "/mnt/workspace/src" in matches
+    assert "/mnt/workspace/node_modules" not in matches
+    assert "/mnt/workspace/node_modules/lib" not in matches
+    assert truncated is False
+
+
+def test_aio_sandbox_grep_invalid_regex_raises() -> None:
+    with patch("deerflow.community.aio_sandbox.aio_sandbox.AioSandboxClient"):
+        sandbox = AioSandbox(id="test-sandbox", base_url="http://localhost:8080")
+
+    import re
+
+    try:
+        sandbox.grep("/mnt/workspace", "[invalid")
+        assert False, "Expected re.error"
+    except re.error:
+        pass
+
+
+def test_aio_sandbox_glob_parses_json(monkeypatch) -> None:
+    with patch("deerflow.community.aio_sandbox.aio_sandbox.AioSandboxClient"):
+        sandbox = AioSandbox(id="test-sandbox", base_url="http://localhost:8080")
+    monkeypatch.setattr(
+        sandbox._client.file,
+        "find_files",
+        lambda **kwargs: SimpleNamespace(data=SimpleNamespace(files=["/mnt/user-data/workspace/app.py", "/mnt/user-data/workspace/node_modules/skip.py"])),
+    )
+
+    matches, truncated = sandbox.glob("/mnt/user-data/workspace", "**/*.py")
+
+    assert matches == ["/mnt/user-data/workspace/app.py"]
+    assert truncated is False
+
+
+def test_aio_sandbox_grep_parses_json(monkeypatch) -> None:
+    with patch("deerflow.community.aio_sandbox.aio_sandbox.AioSandboxClient"):
+        sandbox = AioSandbox(id="test-sandbox", base_url="http://localhost:8080")
+    monkeypatch.setattr(
+        sandbox._client.file,
+        "list_path",
+        lambda **kwargs: SimpleNamespace(
+            data=SimpleNamespace(
+                files=[
+                    SimpleNamespace(
+                        name="app.py",
+                        path="/mnt/user-data/workspace/app.py",
+                        is_directory=False,
+                    )
+                ]
+            )
+        ),
+    )
+    monkeypatch.setattr(
+        sandbox._client.file,
+        "search_in_file",
+        lambda **kwargs: SimpleNamespace(data=SimpleNamespace(line_numbers=[7], matches=["TODO = True"])),
+    )
+
+    matches, truncated = sandbox.grep("/mnt/user-data/workspace", "TODO")
+
+    assert matches == [GrepMatch(path="/mnt/user-data/workspace/app.py", line_number=7, line="TODO = True")]
+    assert truncated is False
+
+
+def test_find_glob_matches_raises_not_a_directory(tmp_path) -> None:
+    file_path = tmp_path / "file.txt"
+    file_path.write_text("x\n", encoding="utf-8")
+
+    try:
+        find_glob_matches(file_path, "**/*.py")
+        assert False, "Expected NotADirectoryError"
+    except NotADirectoryError:
+        pass
+
+
+def test_find_grep_matches_raises_not_a_directory(tmp_path) -> None:
+    file_path = tmp_path / "file.txt"
+    file_path.write_text("TODO\n", encoding="utf-8")
+
+    try:
+        find_grep_matches(file_path, "TODO")
+        assert False, "Expected NotADirectoryError"
+    except NotADirectoryError:
+        pass
+
+
+def test_find_grep_matches_skips_symlink_outside_root(tmp_path) -> None:
+    workspace = tmp_path / "workspace"
+    workspace.mkdir()
+    outside = tmp_path / "outside.txt"
+    outside.write_text("TODO outside\n", encoding="utf-8")
+    (workspace / "outside-link.txt").symlink_to(outside)
+
+    matches, truncated = find_grep_matches(workspace, "TODO")
+
+    assert matches == []
+    assert truncated is False
+
+
+def test_glob_tool_honors_smaller_requested_max_results(tmp_path, monkeypatch) -> None:
+    runtime = _make_runtime(tmp_path)
+    workspace = tmp_path / "workspace"
+    (workspace / "a.py").write_text("print('a')\n", encoding="utf-8")
+    (workspace / "b.py").write_text("print('b')\n", encoding="utf-8")
+    (workspace / "c.py").write_text("print('c')\n", encoding="utf-8")
+
+    monkeypatch.setattr("deerflow.sandbox.tools.ensure_sandbox_initialized", lambda runtime: LocalSandbox(id="local"))
+    monkeypatch.setattr(
+        "deerflow.sandbox.tools.get_app_config",
+        lambda: SimpleNamespace(get_tool_config=lambda name: SimpleNamespace(model_extra={"max_results": 50})),
+    )
+
+    result = glob_tool.func(
+        runtime=runtime,
+        description="limit glob matches",
+        pattern="**/*.py",
+        path="/mnt/user-data/workspace",
+        max_results=2,
+    )
+
+    assert "Found 2 paths under /mnt/user-data/workspace (showing first 2)" in result
+    assert "Results truncated." in result
+
+
+def test_aio_sandbox_glob_include_dirs_enforces_root_boundary(monkeypatch) -> None:
+    with patch("deerflow.community.aio_sandbox.aio_sandbox.AioSandboxClient"):
+        sandbox = AioSandbox(id="test-sandbox", base_url="http://localhost:8080")
+    monkeypatch.setattr(
+        sandbox._client.file,
+        "list_path",
+        lambda **kwargs: SimpleNamespace(
+            data=SimpleNamespace(
+                files=[
+                    SimpleNamespace(name="src", path="/mnt/workspace/src"),
+                    SimpleNamespace(name="src2", path="/mnt/workspace2/src2"),
+                ]
+            )
+        ),
+    )
+
+    matches, truncated = sandbox.glob("/mnt/workspace", "**", include_dirs=True)
+
+    assert matches == ["/mnt/workspace/src"]
+    assert truncated is False
+
+
+def test_aio_sandbox_grep_skips_mismatched_line_number_payloads(monkeypatch) -> None:
+    with patch("deerflow.community.aio_sandbox.aio_sandbox.AioSandboxClient"):
+        sandbox = AioSandbox(id="test-sandbox", base_url="http://localhost:8080")
+    monkeypatch.setattr(
+        sandbox._client.file,
+        "list_path",
+        lambda **kwargs: SimpleNamespace(
+            data=SimpleNamespace(
+                files=[
+                    SimpleNamespace(
+                        name="app.py",
+                        path="/mnt/user-data/workspace/app.py",
+                        is_directory=False,
+                    )
+                ]
+            )
+        ),
+    )
+    monkeypatch.setattr(
+        sandbox._client.file,
+        "search_in_file",
+        lambda **kwargs: SimpleNamespace(data=SimpleNamespace(line_numbers=[7], matches=["TODO = True", "extra"])),
+    )
+
+    matches, truncated = sandbox.grep("/mnt/user-data/workspace", "TODO")
+
+    assert matches == [GrepMatch(path="/mnt/user-data/workspace/app.py", line_number=7, line="TODO = True")]
+    assert truncated is False

config.example.yaml

@@ -325,6 +325,16 @@ tools:
     group: file:read
     use: deerflow.sandbox.tools:read_file_tool
 
+  - name: glob
+    group: file:read
+    use: deerflow.sandbox.tools:glob_tool
+    max_results: 200
+
+  - name: grep
+    group: file:read
+    use: deerflow.sandbox.tools:grep_tool
+    max_results: 100
+
   - name: write_file
     group: file:write
     use: deerflow.sandbox.tools:write_file_tool