Pexo

THE PEXO PROMPT LIBRARY / TEMPLATE

ChatGPT Prompt Template: Role, Context and Output Format

Write a structured ChatGPT prompt from your own task details. The page gives a fill-in template and examples; it does not run ChatGPT or generate prompt text automatically. Fill the worksheet, copy the completed prompt into the relevant workflow, and review the actual result. The examples below are editorial drafts, not observed outputs.

A structured prompt worksheet made of three blank physical cards for role, context and output, arranged with precise spacing beside a pencil and an unbranded notebook
Illustrative concept image made for this page · not a tested product output.

ChatGPT Prompt Worksheet

Replace each bracket with your own information. Keep unknown facts out of the brief.

Prompt text
Role: [relevant task perspective]
Task: [one concrete job]
Source context: [approved facts or files]
Boundaries: [what not to assume or change]
Output format: [list, table, shot plan, etc.]
Review: [facts, wording and omissions to check]

Which brief fits your source?

Brief 1: Role-task-source worksheet. Brief 2: Three-scene product-video planner.

Selection rule: Use the worksheet to structure any approved source; use the product example when a reviewer can verify each scene's claim.

Source-to-instruction relationship: Source facts and output format are separate: the assistant can plan scenes but cannot supply missing product benefits.

If the draft misses the brief

Visible signal: A scene asserts an unapproved price or benefit because the role sounds authoritative.

Next revision: Add a verification note beside the claim and ask for its source before keeping that scene.

Keep the source and review together

Source to retain: the approved product facts and the completed role-task-source worksheet. After an actual run: save the exact prompt version, resulting file and a review note stating whether every proposed scene's product claim has a matching fact and missing claims remain flagged. Until that run exists, leave the result and review fields unfilled. The Gallery cases below are separate published projects and cannot complete this record.

Set Role, Context and Output

Give the task before a long style instruction. Put source facts in a separate block and tell the assistant to flag missing information. Choose an output format that lets you review the answer.

Copyable Task Templates

Example 1: Task: plan a product-video script from approved claims.

Prompt text
You are a careful video planner. Use only the approved product facts I provide. Draft three scenes with visual action, narration and a verification note for each claim. If a fact is missing, ask for it; do not invent a price or benefit.

Example 2: Task: summarize a research note for an internal meeting.

Prompt text
You are an editor. Summarize the supplied note in five bullets, keeping dates and source names exact. Separate stated facts from your inferences. End with two questions that need confirmation.

These are written example prompts; they are not screenshots or measured outputs.

Worked example: turn approved notes into a reviewable task

Suppose the only supplied facts are a blue travel mug, a screw lid visible in two approved photos, and no verified heat-retention claim. The audience is an online shopper. Fill the worksheet with those facts rather than asking for a generic “compelling ad.”

Prompt text
Role: Careful product-video planner
Task: Draft a three-scene outline for an online shopper
Source context: Approved front and lid photos of a blue travel mug; screw lid is visible
Boundaries: Do not invent insulation claims, a price or unseen product features
Output format: Scene | Visual from source | Narration | Fact to verify
Review: Check lid shape, color and every claim against the source photos and approved copy

The requested columns are an output format, not a sample ChatGPT answer. If a draft mentions “keeps drinks hot,” remove that unsupported line and ask for a scene based only on the visible lid and mug shape.

Check the Result

A well-structured prompt can reduce ambiguity but cannot guarantee factual or safe answers. Review outputs against the underlying source, especially numbers and quotations.

Use a Video Brief in Pexo

If the result is an approved video brief, bring that brief and assets to Pexo's AI video generator and inspect the planned scenes and preview.

Privacy and Usage Notes

Use only references you have permission to submit. Remove unnecessary personal or customer information. Before sending sensitive material to any service, review its current terms and privacy policy; this worksheet makes no claim about storage, deletion or account conditions.

Inspect the prompt's source boundary

Use this brief when: Use the worksheet for a task with known source facts and a format a reviewer can check.

Acceptance check: Before copying, identify the role, task, approved context, prohibited assumptions and output fields; then compare the response against the source.

If it fails: If a fact is missing, ask for it rather than letting the prompt imply one.

Run a missing-source challenge before reuse

For the role-context-output template, prepare one ordinary input and one with a missing product fact. State the expected response before running either: the assistant should use the supplied context and flag the absent claim rather than filling it in. Save the exact prompt, source excerpt, model/date and two outputs so a reviewer can compare the claimed boundary with observed behavior.

OpenAI's prompt-engineering guide distinguishes instructions from user input and recommends clear boundaries for context. That supports this template structure, but it does not establish that this page ran ChatGPT or that every model follows the boundary. The Pexo Gallery poster interface is only a visual analogy for separating controls; it is not a ChatGPT prompt test.

ChatGPT Prompt FAQ

Does this page generate a ChatGPT response?

No. Copy the template into the chat product you use, then review its response.

Where should source facts go?

Place approved facts in a separate context block and tell the chat product to flag missing information rather than invent it.