Skill Details
Runtime
(claude-5-5-sonnet)
Example user message
[userMessage]: Run a specificity check on this scene before I write prompts: INT. APARTMENT - NIGHT. He puts on a jacket and grabs the keys off the table. He walks out the door.
Example agent response
Here's the audit:
1. "He puts on a jacket" — Vague detail (medium). No color, cut, or material is given, so the generator will invent one and it likely won't match the jacket in other shots of this character. → What color and style is the jacket, and is he mid-way through putting it on or already wearing it?
2. "grabs the keys off the table" — Unstated layering (low). It's not clear what else is on the table or where the keys sit relative to it. → Is there anything else on the table the generator should keep or leave out?
3. "He walks out the door" — Missing negative (low). An open door with no stated exterior can invite the generator to add a street, hallway, or people that aren't part of the scene. → What's visible through the door once it's open?
Verdict: the jacket detail is a medium issue and should be resolved before handoff; the other two are low and can be left to the prompt-writer's judgment if you're ready to move fast.
Get this skill + 9 more with SelectFrom $14/mo · Cancel anytimeFrom $14/mo · Cancel anytime
View SelectBy purchasing this skill, you agree to our terms of service
CLAUDE-5-5-SONNET
Catches vague, under-specified lines in a script before they become AI image/video prompts. A 12-category audit built from real production mistakes: missing details, unstated layering, missing negatives, new characters needing references, face re-description risk, unspecified on-screen text, dialogue issues, and more. For each issue you get the exact quote, a severity, why it's a problem, and one concrete question that fixes it. Works on scripts in any language.
...more
Added 1 week ago
