tapestry_skill

Overview

This skill unifies content extraction and action planning for any URL. It detects the content type (YouTube, article, PDF), extracts the source text or transcript, then automatically generates a Ship-Learn-Next action plan. The workflow saves both the raw content file and a concise, multi-step implementation plan.

How this skill works

The skill first detects URL type using simple pattern checks and HTTP headers. It runs the appropriate extractor: YouTube transcripts via yt-dlp, HTML articles via reader/trafilatura or a fallback parser, and PDFs via download + pdftotext when available. After extraction it runs the Ship-Learn-Next planner to produce a 4–8 week, rep-based action plan and saves both files, then presents a summary and next-step prompt.

When to use it

• You say: "tapestry <URL>", "weave <URL>", or similar intent to extract and plan
• You have a YouTube video and want a transcript plus actionable tasks
• You found an article or blog post you want turned into a repeatable learning quest
• You have a PDF (paper/report) and need extracted text and a practical implementation plan
• You want an end-to-end extract→plan workflow without manual steps

Best practices

• Provide a publicly accessible URL to avoid authentication failures
• Prefer plain article or video links; PDFs should be direct links or return application/pdf
• Review the first 10 lines of extracted content before acting on the plan
• Commit to a specific date for Rep 1 when asked—this improves follow-through
• Install recommended tools (yt-dlp, reader/trafilatura, pdftotext) to maximize extraction quality
• If extraction fails, try alternative extractor or provide the content file directly

Example use cases

• User: "tapestry https://example.com/how-to-build-saas" → extracts article and creates a 4-week build-and-test quest
• User: "weave https://youtube.com/watch?v=..." → downloads transcript and generates video-production practice reps
• User: "help me plan https://example.com/research.pdf" → downloads PDF, extracts text, and creates an applied-research implementation plan
• User provides a blog post URL and asks to "make this actionable" → returns saved content file plus Ship-Learn-Next Plan markdown
• Fallback: user gives a complex site; skill attempts HTML extraction and reports if manual upload is needed