The Forge

Raw trend data goes in. Brand-native, source-traceable content concepts come out.

The Forge runs every input through three psychology frameworks in sequence, but it does NOT jump straight from trend to concept. It first identifies what made the source outlier work, decides which parts of that pattern are portable, translates those parts through the active brand voice, then outputs structured concepts you (or an agent) can take straight to a scheduler or content calendar.

When the upstream Pattern Report is provided as JSON, The Forge normalizes and prioritizes the richer Content DNA fields instead of only reading the markdown summary. That means it can reason from the source video's brick, why-it-worked explanation, and use-this-when guidance, not just the topic and angle.

When to use this skill:

• You have a trending topic, viral post, or emerging idea and need content angles
• You want to convert competitor content into original concepts
• You have raw research from last30days, content-calendar, or similar skills
• You need 3-5 distinct platform-specific concepts fast

What you get back: Numbered concepts, each with a hook, angle, body arc, format, platform fit, source attribution, and visible pattern lineage. When a structured Content DNA Pattern Report is available, The Forge also explicitly uses brick_name, brick_definition, why_it_worked, and plain_english_takeaway to improve transfer quality. Final concepts should surface that logic explicitly, including a source why-it-worked summary and a plain-English Use this when... line.

When iterating or debugging the skill: Read references/lessons.md for the practical learnings from real test runs, especially around hook simplicity, visible transfer, execution notes for experiments, and avoiding invented specifics.