AgentSkillsCN

paper2wechat-pipeline

通过串联解析器、风格证据、文章Markdown路径解析，以及微信发布者等步骤，将从ArXiv输入到微信就绪输出的端到端paper2wechat工作流程进行编排。当用户请求从ArXiv到微信草稿的一站式流程/单一入口执行时使用此功能。

SKILL.md
--- frontmatter
name: paper2wechat-pipeline
description: Orchestrate end-to-end paper2wechat workflow from arXiv input to WeChat-ready outputs by chaining parser, style evidence, article markdown path resolution, and wechat-publisher steps. Use when users ask for 一条龙流程/单入口执行 from arXiv to WeChat draft.

Paper2WeChat Pipeline

Use this skill as the orchestration layer over existing skills:

  • •paper2wechat for parsing + article writing
  • •wechat-publisher for rendering + image upload + draft creation

Inputs

Required:

  • •paper input: arXiv URL / arXiv ID / local PDF path

Optional:

  • •cache root (default .paper2wechat)
  • •article markdown path (--article-md)
  • •publish options (--theme, --upload-images, --create-draft, --auto-thumb)

Command

bash
python .agents/skills/paper2wechat-pipeline/scripts/run_pipeline.py \
  --paper "2602.09013" \
  --theme ai-insight \
  --upload-images

Behavior

  1. •Parse paper and extract figures via paper2wechat/scripts/fetch_paper.sh.
  2. •Run style evidence script via paper2wechat/scripts/detect_style.py.
  3. •Resolve expected article markdown path (.paper2wechat/<paper_id>/outputs/<paper_id>.md).
  4. •If markdown exists, call wechat-publisher/scripts/publish_wechat.py.
  5. •If markdown is missing, report parse/style outputs and stop with actionable guidance.

Notes

  • •This skill is intentionally non-invasive: it does not change parser/writer logic.
  • •For true one-shot draft creation, ensure markdown exists first (by Agent writing step) then enable publish flags.