Generate and edit Draw.io, Mermaid, and Excalidraw diagrams from natural language using a structured JSON spec.
Documents
tibetan-buddhist-product-article-generator
Try it读取 ~/.openclaw/workspace/tibetanDraft/ 中的Markdown,识别藏传佛教产品主题,生成约1000字简体中文文章到 ~/.openclaw/workspace/tibetanProc/。 只写藏传佛教,事实需网络核验,含至少3章节+主图+章节图。文件名用时间戳格式。 Read...
What it does
Tibetan Buddhist Product Article Generator
The skill document
Tibetan Buddhist Product Article Generator
🎯 Purpose
自动生成符合藏传佛教标准的中文产品文章,包含网络事实核验和藏式图像。
📂 Input/Output Paths
输入目录: ~/.openclaw/workspace/tibetanDraft/
输出目录: ~/.openclaw/workspace/tibetanProc/
文件命名规则 (必须严格遵守):
yymmddHHMM_主体名.md
yymmddHHMM_main.png # 主图
yymmddHHMM_section1.png # 章节1图
yymmddHHMM_section2.png # 章节2图
🔄 完整执行流程
1. 初始化检查
检查 ~/.openclaw/workspace/tibetanDraft/ 是否存在 检查是否有 .md 文件 如果目录不存在或为空 → 报错 "草稿目录为空,请先放入输入文件" 如果多个文件 → 读取最新的一个
2. 主题识别
从输入文件提取藏传佛教产品主题,如:唐卡、法器、佛像、经书等。
3. 网络研究 (必须)
搜索查询示例: "Tibetan Buddhism [产品] history verified" "Tibetan [产品] traditional use past present" "Tibetan Buddhism [产品] materials facts"
4. 文章生成规则
- 语言: 口语化、非正式简体中文
- 长度: 约1000字
- 章节 (至少3个):
- 藏族历史与藏传佛教历史相关背景
- 产品信息核验 (材料、制作、用途)
- 藏族过去和现在的用法
- 对拥有者的宗教文化身心价值 (不夸大)
- 事实原则: 未验证的信息写"资料显示"或"未确认"
5. 图像生成
主图提示: "Traditional Tibetan Buddhist [产品] in authentic thangka style, rich colors, gold details, temple setting" 章节图提示: "Tibetan [场景] featuring [产品], authentic religious art style" 输出: PNG, 同目录, 时间戳文件名
6. 输出模板
# [产品名称]
摘要
...
## 第一节:历史背景
...
## 第二节:产品信息
...
(以此类推)
⚠️ 错误处理与回退
| 错误场景 | 处理方式 | 输出文件 |
|---|---|---|
| 草稿目录不存在 | 创建目录并提示 | 无 |
| 无输入文件 | 输出说明文件 | yymmddHHMM_usage.md |
| 主题无法识别 | 列出可能主题,等待确认 | yymmddHHMM_topics.md |
| 网络搜索失败 | 使用本地知识 + 标记"需进一步验证" | 正常输出 + 警告 |
| 图像生成失败 | 输出无图版本 + 记录日志 | yymmddHHMM_log.txt |
🧪 测试用例
输入文件: test_tibetanDraft/thangka_draft.md
我想写一篇关于唐卡的文章...
预期输出:
2604071901_thangka.md (约1000字) 2604071901_main.png 2604071901_section1.png (历史) 2604071901_section2.png (产品信息) ...
📋 元数据文件 (自动生成)
每个输出文件夹会附带:
skill-run.json: { "timestamp": "2604071901", "input_file": "原文件名", "product": "识别的产品", "word_count": 1024, "images_generated": 4, "research_sources": ["web:1", "web:11", ...] }
🚀 自动化运行参数
--dry-run: 只检查不生成 --product=强制指定产品名 --draft=指定输入文件 --output-dir=自定义输出目录
此技能符合 ClawHub 发布标准,已包含完整错误处理、测试用例和自动化模板。
Related skills
Find why your productivity system keeps failing, then apply the smallest fix — capacity math, bottleneck routing, durable local notes.
Stores durable facts in a categorized, plain-markdown vault on disk, alongside your agent's built-in memory.
Join a video meeting as an AI bot with voice, avatar, and screenshare across four operating modes.
Fetch raw ad creative, app, ranking, and revenue data from AdMapix as structured JSON.
Site audit, content writing, and competitor analysis for organic search rankings.
More from j3ffyang
Browse all skillsPolish a technical blog draft into an 800–1000 word, 3–4 section zh-CN article, preserve technical terms/code, and generate consistent hero + per-section image prompts when the user asks to polish and translate a blog with images.
Generate video from supplied image using Veo-3.0 async API. Gemini vision analyzes image, Veo creates video via predictLongRunning. All outputs in ~/.openclaw/workspace/tibetanProc/ with yymmddHHMM prefix.
Read a plain Markdown file (e.g. 260321_openclawConfig.md), generate header and section images using the default OpenClaw image model and API key from ~/openclaw/.env, create a new *_img.md copy with embedded PNGs, and move all generated images into an img/ folder next to the original file.
Polish a technical blog draft into a 1000–1200 word, 4–5 section en-US article, preserve technical terms/code, and generate one consistent hero image prompt.
Help the user draft, review, and polish blog posts in Markdown, keeping the original languages and informal tone.
Generate a daily AI news newsletter from fresh web sources. Use when the user asks for a current AI digest, AI news roundup, curated newsletter, or daily AI...