srt2md

将 SRT 字幕文件转换为结构化 Markdown 文档的完整工作流。无需配置任何额外 API key，完全依赖当前 session 内置能力完成全部处理。 适用场景：用户上传或提供 .srt 字幕文件，想要将视频内容转化为可读文档；整理 Bilibili / YouTube 视频字幕；将字幕内容分析后生成知识文...

Security Scan

The skill's files, instructions, and single Python helper are consistent with its stated purpose (convert SRT → structured Markdown); it requests no credentials and has no surprising install/network behavior in its own code.

Purpose & Capability

Name/description match the actual artifacts: a small script to convert .srt to a cleaned .txt, a markdown template, and instructions to produce a structured .md. There are no unrelated environment variables, binaries, or config paths requested.

Instruction Scope

Runtime instructions stay within the conversion/analysis workflow. Stage 2 explicitly runs analysis in-session (no external APIs). Stage 3 performs web searches and recommends using an external 'multi-search-engine' skill or a built-in 'web_search' fallback; this means the overall workflow can access external search results (expected for enrichment) — the skill itself does not call or store secrets.

Install Mechanism

No install spec; this is instruction-only with a small included Python script. Nothing is downloaded from external URLs or written to non-standard locations by an installer.

Credentials

The skill declares no required environment variables, credentials, or config paths. The included script optionally imports chardet if present but does not require any secrets.

Persistence & Privilege

always:false and normal agent invocation. The skill does not request persistent/global privileges or modify other skills' configuration.

Assessment

This skill appears to do exactly what it says: it runs a local Python script to produce a time-stamped .txt from an .srt and then uses in-session analysis plus optional web search enrichment to fill a Markdown template. Before installing/using: (1) be aware the script will read the provided .srt and write output files next to it, (2) stage 3 may invoke other skills (multi-search-engine or built-in web_search) which will perform network queries — review those skills for their own permissions and credentials, and (3) the script will use the chardet library if available but does not require it. If you are comfortable with those behaviors, the skill is coherent and reasonable to use.

SRT to Doc

输入 / 输出

• 输入：.srt 字幕文件（支持 UTF-8 / GBK / Big5 等编码）
• 输出：
  • {标题}.md — 结构化分析文档（主交付物）
  • {标题}_subtitles.txt — 带时间轴的字幕原文（辅助交付物）
• 命名约定：取视频标题，中文保留，空格→下划线，特殊字符删除
  • 例：大明王朝1566_第一集.md + 大明王朝1566_第一集_subtitles.txt

---

四阶段流程

阶段一：SRT → TXT

python scripts/srt_to_txt.py input.srt {标题}_subtitles.txt

脚本自动检测编码，输出带时间轴的干净 TXT，并打印条数与时长。 TXT 为独立交付物，不进入最终 Markdown。

---

阶段二：内容分析

读取 TXT 后在当前 session 内完成，不调用任何外部 API：

• 识别主题与核心议题，划分内容段落
• 提取 3-7 个核心观点（附时间戳）
• 列出关键概念、人物、专有名词
• 判断内容类型（影响阶段三的搜索方向）：
  • 知识/技术类 → 四个搜索方向全用
  • 人文/叙事类（历史剧、访谈、纪录片）→ 跳过"工具/产品"方向，"最新动态"改搜史料/研究/评论

---

阶段三：搜索补充

详细搜索策略见 references/search_strategy.md。

简要规则：

• 基于阶段二提炼的概念、人物、工具名提炼搜索词
• 优先调用 multi-search-engine skill（若可用）；不可用时用内置 web_search
• 每个关键词取最相关 2-3 条，记录来源 URL

---

阶段四：输出 Markdown

读取 assets/doc_template.md，按模板结构填写内容。

必填章节：内容概要、内容结构、核心观点 条件章节（有内容则写，无则省略）：关键概念、主要人物、相关动态/延伸阅读

---

下游联动

文档生成后可直接交给以下 skill 继续处理：

目标	调用 skill
发布为微信公众号文章	wechat-mp-writer（建议用"历史"或"技术/AI"模板）
搜索同类公众号参考文章	wechat-article-search