#multimodal prompts｜PromptCraft

SAMPLE一段「看到什麼 + 最該注意的 + 3 個下一步」的可執行回覆。

2026-06-04

Gemini 2.5 Pro：丟一張截圖，直接吐出「下一步該做什麼」

用 Gemini 2.5 Pro 多模態，把任何截圖（錯誤訊息 / 後台 / 報表 / 對話）變成具體可執行的下一步清單。

SAMPLE5 段結構化文獻綜述、含 methodology table、引用具體 paper 名稱、誠實標 hedge

Read 10 papers at once and synthesize them with Gemini

Upload 10 PDFs and Gemini automatically maps the consensus, disagreements, and gaps, then drafts a literature review.

SAMPLE4 段分析：business quality / management tone（引言）/ red flags（含 page #）/ 3 個提問

Auto-Brief a 10-K with Gemini

Upload a public company's annual or quarterly report PDF and auto-generate a management-quality + risk read plus an investment thesis

Advanced

SAMPLE4 stage 讀檔計畫、每檔有「look for X」hint、加「不要看這些」防嚇跑

Onboard New Hires to a Codebase with Gemini

Upload the whole repo or its key files, and Gemini gives new hires a reading order to understand the codebase within an hour

Intermediate

SAMPLE3 段：visual diagnosis / 3 個候選修法（含信心度）/ 還需要看什麼確認

Reverse-Engineer a UI Bug from a Screenshot with Gemini

Paste 1-3 screenshots (browser / Figma / app) and Gemini pinpoints the visual problem + the fix

Intermediate

SAMPLE結構化 diff list、每條標 section + page + materiality + 1 句解釋、結尾 executive summary

Compare two contract PDFs and find the differences with Gemini

For legal, procurement, and HR teams: upload two PDFs and get an automatic line-by-line list of clause differences.

SAMPLE結構化視覺批評、引用具體元素、不模糊建議

2026-04-25

Deep multimodal image analysis with Gemini

Upload an image for structured analysis: design reviews, UX audits, and competitive screenshot comparisons.

Intermediate