Bug type

Regression (worked before, now fails)

Beta release blocker

No

Summary

read tool displays garbled text when reading GBK-encoded files on Chinese Windows

Steps to reproduce

1. On Chinese Windows 11 (default encoding GBK/codepage 936)
2. Create a text file with GBK encoding using PowerShell:

   $content = "中文测试内容"
   [System.IO.File]::WriteAllText("test.txt", $content, [System.Text.Encoding]::Default)
   

Use OpenClaw's read tool to read the file
Observe garbled text output (mojibake)

Expected behavior

The read tool should correctly display Chinese text from GBK-encoded files, either by:

• Auto-detecting file encoding, or
• Respecting system default encoding on Windows, or
• Providing an encoding parameter to specify file encoding

Actual behavior

GBK-encoded files are decoded as UTF-8, resulting in garbled text (mojibake).

Example:

• File content (GBK): "GBK 编码测试\n公司：深圳欧盛自动化"
• read tool output: "GBK \n˾ŷʢԶ"

UTF-8 encoded files work correctly.

OpenClaw version

2026.6.1

Operating system

Windows 11 23H2 (Chinese Simplified)

Install method

npm global

Model

qwen/qwen3.5-plus

Provider / routing chain

qwen/qwen3.5-plus → gateway

Additional provider/model setup details

No response

Logs, screenshots, and evidence

### 测试证据

**UTF-8 文件读取正常：**

**GBK 文件读取乱码：**

# 用系统默认编码 (GBK) 写入
[System.IO.File]::WriteAllText("gbk_test.txt", "中文内容", [System.Text.Encoding]::Default)
# 字节检查：前 10 字节 = 47 42 4B 20 B1 E0 C2 EB B2 E2 (GBK 编码)

write → temp/中文测试.txt (UTF-8)
read → ✅ 正常显示中文


**GBK 文件读取乱码：**

# 用系统默认编码 (GBK) 写入
[System.IO.File]::WriteAllText("gbk_test.txt", "中文内容", [System.Text.Encoding]::Default)
# 字节检查：前 10 字节 = 47 42 4B 20 B1 E0 C2 EB B2 E2 (GBK 编码)


read → temp/gbk_test.txt
输出：❌ "GBK \n˾ŷʢԶ"

Impact and severity

Affects Chinese Windows users when reading common GBK-encoded text files.
Workaround: Use exec with PowerShell to read files with explicit encoding, or convert files to UTF-8 first.

Additional information

添加 encoding 参数到 read 工具
或自动检测文件编码（通过 BOM 或内容分析）
或在 Windows 上默认使用系统编码