生成符合中国公文格式规范的Word文档，支持标题、正文样式、自动格式排版和中文引号配对。

This package appears to implement what it claims (formatting .docx documents) and doesn't request secrets or network access, but it contains several practical issues you should address before running it: - Do NOT run install.sh unreviewed. It uses an unknown 'uv' command. Replace with standard commands (for example: python3 -m venv .venv && source .venv/bin/activate && pip install python-docx) or inspect what 'uv' would do on your system. - quick_api.py will fail as written: it imports 'docx_formatter' but the supplied file is named 'docx-formatter.py' (hyphens make it unimportable). Rename files or fix imports before using the API helper. - examples/content.json appears to contain unescaped double quotes inside a JSON string (invalid JSON). Fix/validate the example JSON to avoid runtime json.load errors. - The SKILL.md suggests passing an author with a literal '\n' sequence; the implementation splits on the two-character string '\\n' rather than actual newline characters—test and decide on a consistent convention (or accept real newlines). - Because the skill documents automatic triggers for politically sensitive document types, be mindful of the content you feed into the tool and your local policies for generating or storing such materials. Recommended next steps: run the code in an isolated environment (container or throwaway VM), correct the installer and import issues, validate example files, and run the test.sh only after fixing these problems. If you are not comfortable fixing the repository, treat it as untrusted/broken and avoid executing the install script on a production machine.

Type: OpenClaw Skill Name: docx-formatter Version: 1.0.1 The skill bundle is designed for a legitimate purpose (document formatting) and does not contain explicit malicious code for data exfiltration, backdoors, or unauthorized remote control. However, the `docx-formatter.py` and `quick_api.py` scripts directly use user-provided file paths for `--output`, `--content`, and `--from-markdown` arguments without sanitization. This creates a path traversal vulnerability, allowing an attacker (via prompt injection to the agent) to potentially read or write files to arbitrary locations on the system, which could lead to information disclosure or unauthorized file modification.