skills.rest
Kaleb-Rupe
Kaleb-RupeCOMMUNITY
Explore all skills by this author →
𝕏@MightieMagsGitHub

Harvest

Mine one piece of content for ideas worth adopting into the Aurora system.

Updated Jul 10, 2026
Search Tags:#yt-dlp#system evaluation#knowledge harvesting#content mining#youtube transcript#idea extraction#ranked report
One-click install
npx skills add https://github.com/Kaleb-Rupe/aurora --skill harvest-kaleb-rupe
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: Harvest
Source: https://github.com/Kaleb-Rupe/aurora/tree/main/claude/skills/Harvest
Command: npx skills add https://github.com/Kaleb-Rupe/aurora --skill harvest-kaleb-rupe

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires yt-dlp.

What problem does it solve? When you encounter a useful URL, YouTube video, article, or text snippet, it is hard to systematically decide which ideas are worth adopting into your existing system. Harvest evaluates a single piece of content against Aurora's actual components and produces a ranked, honest report of what is worth taking. ## Core Features & Use Cases - Multi-source ingestion: Accepts YouTube links (via yt-dlp auto-generated transcripts), article URLs (via WebFetch), local files, or raw pasted text. - Surface mapping: Maps every candidate idea to a concrete Aurora surface such as an Algorithm phase, hook, skill, auto-memory, mind vault, agent, or doctrine file. - Honest status tagging: Tags each candidate NEW, PARTIAL, DONE, or REJECTED by cross-checking the auto-memory ledger, then ranks by usefulness, novelty, and effort. - Use Case: You find a blog post about agent memory techniques. Run Harvest on the URL to get a ranked table showing which techniques Aurora already has, which were previously rejected, and which two are genuinely new and worth proposing. ## Quick Start Ask the assistant to harvest a specific URL or pasted text, for example by saying harvest this article followed by the link, to receive a ranked report of adoptable ideas.

Frequently Asked Questions about Harvest

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I extract useful ideas from a YouTube video for my system?▼

Provide the YouTube URL and the skill fetches the auto-generated transcript using yt-dlp with the --write-auto-subs flag, then analyzes the spoken text. It never scrapes the page HTML, and if English subtitles are unavailable it falls back to other languages or human-authored captions.

How do I evaluate whether a blog article has ideas worth adopting?▼

Pass the article URL and the skill fetches the body with WebFetch, extracts concrete candidates like techniques, tools, and prompts, then maps each to a specific system surface. The output is a ranked table scored by usefulness, novelty, and effort.

Does the Harvest skill modify my system files automatically?▼

No, Harvest is strictly report-only and never edits the system. It produces a proposal set of candidates, and adopting any item into an Algorithm file, hook, skill, or doctrine requires separate explicit approval.

What input types does content mining support?▼

It accepts four input types: YouTube URLs processed via yt-dlp transcripts, other HTTP URLs fetched as articles, local file paths read directly, and raw pasted text analyzed as-is. Only one input is processed at a time.

Why does the skill check a memory ledger before tagging ideas as new?▼

The auto-memory ledger records settled decisions including features already built and directions explicitly declined. Cross-checking prevents re-pitching solved or rejected work, so candidates matching ledger entries are tagged DONE or REJECTED with a citation.

Related Skills

Explore More Software Engineering Skills→
SahilFruitwalaSahilFruitwala

topic-finder

Fetches and dedupes niche headline candidates from RSS feeds and web search for ranking.

Community
Intermediate
Tx1207Tx1207
5

~apply-candidate

Review, preview, approve, and apply skill or preference candidates with audit logs.

Community
Intermediate
NachoFLizaurNachoFLizaur
3

deep-research

Plan and execute multi-source research with search, fetching, and synthesis.

Community
Advanced
rdleclercrdleclerc

uberassess

Generate source-grounded adoption recommendations for ideas and sources.

Community
Intermediate