pitch-visual-promptinglisted

# Pitch Visual Prompting ## Objective Turn a pitch brief into a model-ready visual prompt that makes the strategic idea visible. Keep the workflow brand-agnostic: derive every color, tone, audience, product, IP constraint, and visual convention from the current brief and references. Do not hardcode any past client, campaign, palette, platform, or IP into the output. ## Core Workflow 1. Determine the requested deliverable before writing the prompt. 2. Extract the communication goal, audience, required evidence, brand variables, reference roles, immutable content, and prohibited content. 3. Decide whether each supplied image must be loaded into the image model, inspected only, or used only as an information source. 4. Choose one output mode and one primary composition pattern. 5. Map each important claim or action to something visible in the image. 6. Assemble a concise prompt from only the relevant modules. 7. Check reference fidelity, text risk, hierarchy, realism, and unwanted additions. 8. For revisions, freeze correct regions and describe only the required delta. Read [intake-and-routing.md](references/intake-and-routing.md) when the task type, reference roles, or output mode is unclear. ## Select an Output Mode - **Full-slide image**: Generate the complete pitch page, including layout, copy, and embedded visual evidence. - **Deck visual asset**: Generate a photo, render, collage, product shot, or scene for placement in an editable deck. Default to no visible text.