中文公文写作
PassAudited by VirusTotal on May 12, 2026.
Overview
Type: OpenClaw Skill Name: chinese-official-writing Version: 1.2.15 The skill bundle is a specialized tool for drafting and reviewing Chinese official documents (Gongwen). It consists of comprehensive markdown-based instructions for the AI agent and a Python utility script, `prose_lint.py`, which performs static regex-based analysis on text and DOCX files to identify informal or 'AI-flavored' phrasing. No evidence of data exfiltration, malicious execution, or prompt injection attacks was found; the logic is strictly focused on document structure, tone, and compliance with official standards (e.g., GB/T 9704-2012).
Findings (0)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If a user installs from the live GitHub repository instead of the reviewed ClawHub package or a pinned release, they may receive content that differs from the reviewed artifacts.
The documentation provides user-directed installation prompts that fetch skill directories from GitHub without pinning a commit or tag.
请从 GitHub 仓库 https://github.com/gongyu0918-debug/chinese-official-writing-skill 拉取 ... 目录，并将其安装...
Install the reviewed ClawHub version when possible, or pin and inspect a specific GitHub commit before installing from the repository.
Running the lint script gives local code access to the selected draft files; this is expected for document linting but still requires trust in the script.
The skill documents optional local Python execution over user draft files, including DOCX files.
python .\chinese-official-writing\scripts\prose_lint.py .\draft.docx --structure --format
Run the script only from a trusted reviewed version of the skill, and use copies of important Word documents when testing.
Official work drafts may contain sensitive internal, financial, policy, or procurement information, so any separate reviewer should be chosen carefully.
The workflow allows an independent review step that may involve sharing excerpts or a final draft with another reviewer, although no specific external endpoint or automatic sharing is shown.
Use an independent reviewer only when the user asks or the document is high-stakes... Pass compact excerpts or the final draft, not internal reasoning.
Before using an independent reviewer, confirm whether it is another model, tool, or person, and share only approved excerpts when the document is sensitive.