Week 5.2: Repurposing Content (Long Form & Blog)

Week 5.2 Resources: Repurposing Content (Long Form & Blog)

Class Prompts

You can copy and paste these prompts directly into the Claude Code input bar to replicate the lesson. These commands build upon the project structure established in Week 5.1.

1. In-Depth Research Agent

Create the research workflow file:

CREATE workflows/get_in_depth_research.md

Execute the research workflow:

execute workflows/get_in_depth_research.md
$project_name=mermaids

2. Blog Post Generation

Create the blog post workflow file:

CREATE workflows/create_blog_post.md

Execute the blog post workflow:

execute workflows/create_blog_post.md
$project_name=mermaids

3. Long Form Video Generation

Duplicate the short-form workflow: Note: In the video, we manually copy the folder. You can also use this command.

cp -r workflows/create_short_form_video workflows/create_long_form_video

Rename internal files: Note: You will need to rename the files inside the new folder from “short_form” to “long_form” manually or via command line.

Update the text asset generation logic:

UPDATE workflows/create_long_form_video/create_text_assets.md
- change input to read research files instead of chosen topics
- update prompt to write a long form script (approx 5x longer than short form)
- update segment length logic to 3-4 lines

Update image generation aspect ratio:

UPDATE workflows/create_long_form_video/generate_images.md
- change aspect ratio to 16:9

Execute the long form video workflow:

execute workflows/create_long_form_video/create_long_form_video.md
$project_name=mermaids

4. Troubleshooting & Debugging

Investigate stuck processes:

what happened? only answer don't take any other action

Force a specific fix (Ken Burns Effect):

create tools/video_tools.py based on the two script files. test to make sure it works. mirror the format from tools/ai_image_tools.py. update docs/tools.md. update workflows/create_long_form_video/assemble_video.md

Clean up unnecessary notes in workflow:

remove the unnecessary notes

Key Concepts

• Single Source of Truth: Ensuring all variations (blogs, videos) pull from a central data source rather than referencing each other to avoid circular dependencies.
• Parallel Sub-Agents: Using multiple agents simultaneously to speed up heavy research tasks.
• Context Rot: Understanding when an AI conversation has become too long or complex, leading to degraded performance or “gaslighting,” and how to fix it by starting fresh or creating specific tools.
• The Ken Burns Effect: A panning and zooming effect used on still images to create motion in video, implemented here via Python scripts and FFmpeg.

Required Tools (New for this Week)

• FFmpeg: While used previously, this week relies heavily on complex FFmpeg filters for the Ken Burns effect. Ensure it is installed and added to your system PATH.
  • Download FFmpeg 

Source Code

The complete source code for Week 5 (including the tools, workflows, and .claude folders) is available for download.

• Download Week 5 Files

After downloading, extract the zip file and create a .env file with your API keys.