Three commands. That's the full install. Clone the repo, step inside, and Claude picks up the workspace on its own. No config files to edit, no environment variables to set.

Cinematic prompt methodology, as an installable Claude workspace

Most AI image work fails at the brief, not the model. Three lines in. Generic out. People blame the tool.

I shipped a fix.

pushing-creation installs into Claude. Drop reference images into refs/. Run /frames-brainstorm. Claude reads your refs, runs a DP-style interview, and writes the style pack live as you answer. /frames-shotlist drafts the full storyboard. /frames-shot polishes individual frames.

The brainstorm is where craft enters the pipeline. Claude reads your reference images, then asks the questions a director of photography would ask on a location scout. Your answers become the brief.

Output is markdown. Drop it into PUSHING FRAMES, Midjourney, Sora, or any tool.

Every project carries three artefacts. style.md holds the visual voice, locked down to lens choice, colour science, and negative space rules. storyboard.md is the shot table. refs/ holds the images that seeded everything.

→ github.com/PUSHINGSQUARES/pushing-creation

pushing-creation handles the methodology. PUSHING FRAMES handles generation. They're siblings, not dependencies. You can run either one standalone, or chain them for a continuous pipeline from brainstorm to final render.

You don't get better output by trying harder at the prompt. You get better briefs by treating the model like a director of photography. Specificity transfers craft.

Stop prompting. Start defining outcomes.

//