lieflat-gongwen
通过102万字语料提炼的公文写作 skill，把公文写作风格变成可测量、可验收的量化数据
lieflat-gongwen is an agent skill that enables AI assistants to produce authentic Chinese official government documents (公文) by applying quantified writing style parameters derived from analyzing 1.02 million characters of real administrative writing. The skill supports seven distinct document types including research reports, leadership speeches, work plans, and party-building materials, translating writing style into measurable metrics like sentence length, punctuation density, and heading structure. Compatible with agents.md-standard agents like Codex, ClaudeCode, and Moxt, it uses a six-step workflow that checks document type, applies DNA parameters, generates structure, and validates output against statistical benchmarks.
Key Features
Use Cases
- 01Drafting Chinese government work plans and implementation schemes for administrative departments
- 02Generating research reports analyzing policy effectiveness or industry conditions
- 03Writing leadership speeches and official remarks for government meetings
- 04Creating party-building experience materials and grassroots organizational summaries
- 05Producing work opinions and regulatory deployment documents with proper bureaucratic tone
- 06Training AI agents to distinguish between seven official Chinese document genres based on quantified linguistic fingerprints
Related Skills
View moresuperpowers
An agentic skills framework & software development methodology that works.
skills
Skills for Real Engineers. Straight from my .agents directory.
skills
Public repository for Agent Skills
ponytail
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
lieflat-gongwen — FAQ
What is lieflat-gongwen and what does it do?+
lieflat-gongwen is an agent skill that teaches AI assistants to write authentic Chinese government documents (公文) using statistically-derived style parameters from over 1 million characters of real administrative writing. It covers seven document types including research reports, work plans, leadership speeches, and party-building materials.
How do I install this skill for my AI agent?+
Clone the repository to your agent's skills directory (e.g., `~/.claude/skills/lieflat-gongwen` for Claude-based agents) or place the entire directory in your agent's skills folder. The skill requires zero external dependencies; the optional self-check script uses only Python standard library.
Which AI clients and agents work with this skill?+
The skill is compatible with any agent that supports the agents.md standard, including Codex, ClaudeCode, and Moxt. The author recommends using it in Moxt for its workspace context capabilities when handling long-form document generation.
Do I need API keys or external data to use this skill?+
No API keys are required. The skill includes all necessary parameter cards, DNA documents, and 22 synthetic reference samples. Original source materials are not included due to copyright, but the provided DNA documents and statistical metadata contain all quantified parameters needed.
Is lieflat-gongwen free to use?+
Yes for non-commercial use under the PolyForm Noncommercial 1.0.0 license, including personal research, learning, education, and government use. Commercial use (paid writing services, consulting, commercial product integration) requires separate written authorization from the author.
What makes this different from template-based Chinese official writing tools?+
Instead of providing fill-in-the-blank templates, this skill uses full-corpus statistical analysis to extract measurable linguistic fingerprints (sentence length, punctuation patterns, heading structures). It validates that common template assumptions (like mandatory formulaic openings) actually appear in less than 8% of high-quality real documents.
How do I install lieflat-gongwen?+
Open the source repository on GitHub and follow its README. lieflat-gongwen is a skill — MCP Agents Market links you directly to the official repo.
Is lieflat-gongwen free?+
lieflat-gongwen is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.