Gemini prompts for comparing files and images, building source-bound briefs and planning from real constraints.
Check whether a visual proposal matches written requirements.
Compare FILE and IMAGE for {{decision}}. Return a table with requirement, file evidence, visible image evidence, status and open question. Use only these inputs. Mark [NOT SHOWN] when the image cannot prove a requirement and [NOT IN FILE] when the requirement is absent.Decision: design review. File: CTA must be visible without scrolling. Image: mobile screenshot with CTA below the fold.
Requirement: CTA visible without scrolling. File evidence: explicit requirement. Image evidence: CTA not visible in supplied viewport. Status: does not match.
Before you use itOpen both inputs and reproduce every row before acting on the comparison.
Turn a small approved file set into claims, evidence and open questions.
Create a brief for {{audience}} from these named files: {{files}}. Separate supported claims, conflicts and unanswered questions. For every claim, name the file and relevant section. Do not use outside facts or resolve a conflict without evidence.Audience: launch team. Files: project brief and QA report. Decision: whether blockers remain.
Supported: checkout test passed — QA report, Results. Conflict: launch date differs between brief and QA header. Open question: which date is current?
Before you use itVerify each file and section reference, especially dates, totals and status labels.
Turn tasks, fixed commitments and available time into a reviewable plan.
Build a plan for {{period}} from TASKS, AVAILABLE TIME and FIXED COMMITMENTS. Keep estimates unchanged, respect dependencies, leave {{buffer}} as buffer and list work that does not fit. Do not invent deadlines or assume overtime.Period: this week. Tasks: research 4h, draft 3h. Available: Mon 3h, Wed 4h. Buffer: 1h total.
Mon: research 3h. Wed: research 1h, draft 2h, buffer 1h. Does not fit: 1h draft.
Before you use itRecalculate the hours and check that fixed commitments were not overwritten.
Copy a prompt, replace the text in double braces with your details, then adjust the first draft.
Use labels such as BRIEF, SCREENSHOT and DECISION. Clear labels make it easier to see when the answer blends sources or relies on evidence that is not present.
An image can show a visible element but not confirm an analytics result. A document can state a requirement but not prove implementation. Make that boundary part of the requested output.
Use these prompts when Gemini must work from named text, file or image inputs and expose the evidence behind its answer.
Task: {{task}}. Inputs: {{inputs}}. Decision: {{decision}}. Use only the supplied inputs. Return {{output_format}} with source labels, visible conflicts and [NEEDS INPUT] for missing evidence. End with the three claims most important to verify.Task: compare a brief with a screenshot. Decision: ready for review. Output: requirement evidence table.
Requirement: primary action visible. Brief: required. Screenshot: not visible in supplied viewport. Status: [NEEDS REVIEW].