#📘 User Manual: The Harvester (Bible Edition)

#Introduction

The Harvester is an autonomous research engine designed to find high-demand, low-competition video opportunities for PC Software software. It automates 90% of a researcher's job.

#Input Method 1

(Stream A: The Competitor Vacuum)

This input method focuses on monitoring specific YouTube channels. It acts like a vacuum cleaner, sucking up every single video from a competitor's backlog.

#How it Works

1. Queue a Channel: You give the system a channel handle (e.g., @NotionGuide).
2. Scraping: The system visits the channel's "Uploads" playlist.
3. Smart Resume: It checks 50 videos at a time. If it sees a video date that is already in the database, it stops. This means it has "caught up" to history.
4. Filtering: It ignores Shorts, Livestreams, and videos longer than 60 minutes.

#When to use this?

Use this when you find a competitor who is posting exactly the kind of content you want to make. You want to see everything they have done to find gaps or better angles.

#Input Method 2

(Stream B: The Keyword Harvester)

This input method finds search demand. It asks YouTube Autocomplete "What are people typing?" to discover hidden needs.

#Search Strategies

1. ROOT: Searches just the app name (e.g., "Obsidian").
2. ALPHABET: Searches "Obsidian a", "Obsidian b" ... "Obsidian z". This triggers the "A-Z" autocomplete logic.
3. HOW_TO: Searches "How to Obsidian a" ... "How to Obsidian z".
4. QUESTIONS: Searches "Why Obsidian...", "Can Obsidian...".

#When to use this?

Use this to map the "Mind of the Market". This is how you find topics that no one has made a video about yet (Blue Oceans).

#The Gatekeeper

(Module 2: The AI Filter)

The Gatekeeper is the Brain of the system. It uses a local Llama 3 AI model to review every single video found by Input Methods 1 & 2.

#The 8-Step Pipeline

The AI checks each video against 8 strict rules. If it fails any rule, it is discarded.

1. Format Sorter: Must be a "How-To" Tutorial. No Vlogs, no News, no Reviews.
2. Domain Filter: Must be PC Software. No Mobile Apps, no Hardware, no Consoles.
3. Safety Audit: Scans for "Banned Words" (cracks, hacks, scams).
4. Identity Check: Checks context. "Resolve" must mean "DaVinci Resolve", not "Conflict Resolution".
5. Feasibility: Can a Virtual Assistant make this? (No credit card, no webcam required).
6. Rewrite: Standardizes the title to a clean format: "How to [Action] in [App]".
7. De-Duplication: Checks if we already have this topic in the database.
8. Finalization: Marks the item as READY_FOR_SNIPER.

#The Sniper

(Module 3: Intelligent Validation Engine)

The Sniper is an automated intelligence engine that validates keyword opportunities using mathematical strategies instead of guesswork. It measures velocity (views/day), competition (channel size), and identifies content gaps to give you data-driven verdicts on which keywords are worth pursuing.

#Core Philosophy

"Don't guess. Measure velocity."

The Sniper analyzes the Top 10 search results for each keyword and applies configurable "Guru Strategies" (Niche Hunter, Giant Slayer, Trend Surfer) to determine opportunity scores.

#Key Features

• Proxy Management UI: Visual interface to manage multiple proxy lists with hot-reload support
• War Room Dashboard: Review validated keywords with snapshot previews and score breakdowns
• Strategy Configuration: Customize scoring curves and thresholds for different content approaches
• Operational Security: Pre-flight checks, proxy anonymity validation, and automatic IP protection
• Score-Based Verdicts: Color-coded scores (Green/Yellow/Red) with special flags (Dead Topic ☠️, Dominant Goliath 👑)

#Documentation

For comprehensive guides, see:

• Module 3 Sniper User Guide - For content strategists
• Module 3 Sniper Developer Manual - For developers

#Quick Start

1. Navigate to Proxy Management (/sniper/proxies)
2. Create and activate a proxy list
3. Go to Safety Manual (/sniper/safety) and verify anonymity
4. Launch War Room (/sniper/dashboard) to review validated keywords
5. Approve (✓) high-scoring opportunities for production

#Production Engine

(Module 4: Content Factory)

Placeholder: This module is currently in development.

The Production Engine will take approved topics and generate:

1. Video Scripts
2. Thumbnail Concepts
3. SEO Metadata (Tags, Description)

#Feature One

(Experimental)

Placeholder: New keyword clustering logic based on semantic similarity.

#Feature Two

(Experimental)

Placeholder: Integration with YouTube Data API v3 for real-time rank tracking.

#System Errors & Glossary

#Error Codes

• 404: Resource not found.
• 500: Internal Server Error (usually logic crash).
• 403: Quota Exceeded (Rotate your API Keys).
• ERR_OLLAMA: AI Service is down. Run ollama serve.

#Glossary