YouTube Production — Full Video Content Production Pipeline

Workflow

Phase 1: Preparation (Performed directly by the orchestrator)

1. Extract from user input:
   • Topic/Keywords: The subject the video will cover
   • Channel Info (optional): Channel tone, subscriber count, existing content direction
   • Constraints (optional): Video length, specific requirements
   • Existing Files (optional): Scripts, briefs, or other materials provided by the user
2. Create the _workspace/ directory at the project root
3. Organize the input and save it as _workspace/00_input.md
4. If existing files are provided, copy them to _workspace/ and skip the corresponding phase
5. Determine the execution mode based on request scope (see "Scope-Based Modes" below)

Phase 2: Team Assembly and Execution

Assemble the team and assign tasks. Task dependencies are as follows:

Tasks 2a (script) and 2b (thumbnail) run in parallel. Both depend only on Task 1 (strategy), so they can start simultaneously.

Inter-agent communication flow:

• strategist complete → deliver core angle & tone to writer, title candidates & emotional triggers to designer, keyword map to seo
• writer complete → deliver hook core message to designer (thumbnail-hook consistency), deliver script to seo
• seo complete → deliver title-thumbnail combination feedback to designer
• reviewer cross-validates all deliverables. On 🔴 Must Fix: request revision from the responsible agent → rework → re-validate (up to 2 rounds)

Phase 3: Integration and Final Deliverables

Compile the final deliverables based on the reviewer's report:

1. Verify all files in _workspace/
2. Confirm all 🔴 Must Fix items from the review report have been addressed
3. Report the final summary to the user:
   • Strategy Brief — 01_strategist_brief.md
   • Video Script — 02_scriptwriter_script.md
   • Thumbnail Concept + Images — 03_thumbnail_concept.md
   • SEO Package — 04_seo_package.md
   • Review Report — 05_review_report.md
   • Subtitle File — subtitle.srt

Scope-Based Modes

Adjust the agents deployed based on the scope of the user's request:

Using Existing Files: If the user provides a script, brief, or other existing file, copy it to the appropriate numbered position in _workspace/ and skip that phase's agent. Example: existing script provided → copy to _workspace/02_scriptwriter_script.md → skip writer, deploy only seo, designer, and reviewer.

Data Transfer Protocol

File naming convention: {order}_{agent}_{deliverable}.{extension}

Error Handling

Agent Extension Skills

Agents reference their respective extension skills during task execution to enhance domain expertise.

Test Scenarios

Happy Path

Prompt: "Plan a 10-minute YouTube video as a beginner's guide to AI prompt engineering" Expected Results:

• Strategy brief: 3+ competitive analyses, keyword map, 3 title candidates
• Script: 10-minute length (~1,500 words in English), hook + 3–4 segments + closing
• Thumbnail: A/B concept options + image generation attempt
• SEO: Title/description/tags/chapters/SRT
• Review: All consistency matrix items verified

Existing File Flow

Prompt: "Create SEO optimization and a thumbnail for this script" + script file attached Expected Results:

• Existing script copied to _workspace/02_scriptwriter_script.md
• SEO Mode + Thumbnail Mode merged: seo + designer + reviewer deployed
• strategist, writer skipped

Error Flow

Prompt: "Just write a YouTube script quickly, any topic" Expected Results:

• Switch to Script Mode (strategist + writer + reviewer)
• Topic is unclear, so strategist proposes 3 trending topics before proceeding
• Review report notes "thumbnail/SEO not generated"