Discover and explore top open-source AI tools and projects—updated daily.
dongbeixiaohuoAutomated writing pipeline for humanized content
Top 85.9% on SourcePulse
<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.>
dongbeixiaohuo/writing-agent is a full-stack, "de-AI-fied" writing system using Claude Code, producing nuanced, human-like long-form content. It injects human viewpoints and details via a multi-stage workflow, supporting cost-effective domestic LLMs (DeepSeek, Zhipu GLM, MiniMax). It offers a controllable, reviewable production line from topic selection to publication-ready drafts and WeChat HTML. Targets users needing stable, high-quality, reproducible writing processes beyond one-shot AI generation.
How It Works
This project reframes AI writing as a "production line" with distinct stages: theme selection, stance, "scar tissue" gathering, outline, opening hook generation, draft iteration, review, reader testing, and final output. This pipeline emphasizes control, interruptibility, and reviewability. Leveraging Claude Code, it supports flexible LLM integration (DeepSeek-V3.2, Zhipu GLM, MiniMax) for cost-effective workflow execution.
Quick Start & Requirements
Requires Node.js 18+ and Claude Code CLI (installable via native scripts, Homebrew, or WinGet). Clone the repo, configure LLM API keys (DeepSeek, Zhipu, MiniMax) or use a Claude account. Initiate via claude from the project root. A plugin installation method is also available. No explicit GPU/CUDA requirements listed.
Highlighted Details
Maintenance & Community
Information regarding notable contributors, sponsorships, or community channels (Discord, Slack) is not explicitly detailed in the provided README.
Licensing & Compatibility
The README does not specify the software license, requiring further investigation for adoption decisions, particularly concerning commercial use.
Limitations & Caveats
Not for rapid, one-shot text generation or users uninterested in structured writing processes. Requires user engagement in planning and review. Extensive documentation and multi-stage nature present a learning curve for new users.
3 days ago
Inactive