A Claude Skill that picks your fight before you write a word of copy.
Marketing has no discipline around choosing an opponent. Brands attack whoever the sales team complained about on Monday, ship a comparison page, and then wonder why it converted nothing or, worse, why it introduced a competitor to people who had never heard of them.
Battle rap settled this a long time ago. The cost of a bad matchup is public and it lands the same night. So the culture built real instincts about it. You do not pick a fight because someone annoyed you. You pick it based on what the matchup does to your standing when it is over, win or lose. Position on the ladder. Who owns the crowd. Whether your strengths reach their weaknesses. What they can say back.
That last one is the whole thing. Nobody in marketing writes the competitor's response before writing the attack, and it is the single fastest way to find out that a fight you were excited about is already lost.
versus runs that logic and gives you a verdict. Take it, take it sideways, leave it, or don't say the name.
It stops at the verdict. It does not write the copy. That is on purpose.
Six steps and a scorecard.
It builds the real card first, which usually means telling you that half your competitor list is wrong and that the opponent taking most of your deals is the one nobody put on the slide. Doing nothing. The spreadsheet. The intern. The free tool your buyer already has open.
Then it places you on the ladder honestly, reads whose room you would be fighting in, finds whether your strength reaches something structural in them or just something they can patch by Friday, and writes their comeback in their voice at full strength.
Then it scores it on six dimensions and kills it if it fails any of four rules. Claims you cannot back with a public dated source. A response from them that beats your opening. Punching down. An opponent the room loves.
It will tell you not to take a fight you want. That is most of the value.
git clone https://github.com/MichaelDoesMarketing/versus.git ~/.claude/skills/versusOr grab the packaged .skill file from releases and install it without cloning.
Works in Claude Code, Claude Desktop, and anywhere else skills load.
"We keep losing deals to [competitor]. Should we build a comparison page?"
"Our biggest competitor just raised a Series B. Do we say anything?"
"Who should we actually be positioning against? Here's what we do."
You never have to say the words "battle rap." The skill fires on how people actually talk about this.
Rule around here is that the demo runs on something real, and the most honest test is my own work. So I pointed it at me.
The setup. I sell strategy and messaging as a solo operator. My site says "Your brand isn't boring. It just sounds like everyone else." My whole angle is that I learned persuasion in rooms that gave instant hostile feedback, which is a rep count nobody in a marketing org has.
The fight I wanted was obvious. Go at the positioning-framework world. Frameworks produce correct positioning that still sounds like everything else, because everyone is running the same framework on the same inputs. That is a real argument and it is aimed at a category leader.
What it said.
VERDICT: DON'T SAY THE NAME.
Two kill rules fired at once.
Crowd scored 1. That audience does not merely respect the leading voice in positioning, they learned the discipline from her. Criticism of the method reads to them as criticism of their own competence, and they will defend it harder than she ever would.
Their response scored 5, which is the worst possible score on that dimension because it measures how well they can answer. The comeback writes itself and it is devastating: positioning is a business decision about what you sell and to whom, and it has to be right before anyone writes a memorable line. A great line on wrong positioning fails faster than a boring line on right positioning. That answer does not just beat my opening, it recasts me as the guy who does the fun part after the serious work is finished.
I would have lost that exchange in public and I would not have seen it coming.
Then it found the fight I should be taking.
VERDICT: TAKE IT. Score 24/30.
The opponent is the ghost. Your buyer opens a chat window and gets messaging for free in nine seconds. That is who takes your deals, and it is on nobody's competitive slide.
Ladder 5. Crowd 5, because the room is full of people who already tried it and got mush, and switching cost is zero. Style matchup 5, and structural: a model trained on the average of all marketing writing produces the average of all marketing writing. It cannot stop sounding like everyone else. That is not a bug getting patched, it is what the thing is.
Their response 1. The ghost never answers.
Receipts 3, and that is your homework. You need before and afters from your own work with dates on them.
Then it pointed out that my own headline is already aimed at this opponent and my site frames it as being about competitors. The line was built for the right fight and pointed at the wrong one.
That is the kind of finding I was hoping the skill would produce, and I did not enjoy receiving it.
SKILL.md the workflow, modes, kill rules, output format
references/ladder.md punching up, level, down, and the four opponent types
references/crowd.md whose room it is, loyalty vs shopping, the champion problem
references/style-matchup.md structural vs patchable weakness, the mirror check
references/the-response.md writing their round, and the six standard comebacks
references/scoring.md six dimensions, four kill rules, four verdicts
assets/matchup-scorecard.md fill-in scorecard
assets/fight-card.md the whole division ranked, with you placed honestly
battle-tested writes the round. versus picks the fight. Run versus first.
MIT. Use it, fork it, ship it.