Start here

Paste two or more option blocks plus your priorities into an ordinary AI chat that accepts text. No terminal, API or account setup is needed.

Minimum inputs to paste:

• Options: one block per option, labeled OPTION: <name>, followed by raw specs copied verbatim from a seller, manual, datasheet or your notes.
• Priorities: Criterion: <name> | Weight: <number> | Better direction: higher/lowerexact/yes-no | Acceptable minimum: <value or none>.
• Constraints: budget band, hard caps, must-haves, dealbreakers.
• Decision context in one or two sentences.

Copy the complete prompt that is appended to this page, then paste your inputs where the prompt says "Inputs I will paste".

Check the output by confirming: every number traces to a quoted line you supplied, missing cells read MISSING, every must-have and dealbreaker shows PASS, FAIL or UNKNOWN with a cited line, the arithmetic is visible, and the sensitivity section names the smallest flip condition. The result is a draft for your review, not a purchase decision.

What this prompt does

It restates your decision context, builds a normalized comparison table, checks constraints, shows draft scoring arithmetic, runs a sensitivity pass on the top two options, and lists questions that could resolve each gap. It works only from what you paste; it does not browse, look up products or add values you did not provide.

If your input is empty or only one option is supplied, it should ask for the missing blocks instead of producing a table.

Prerequisites, permissions and limits

• Access to any general-purpose AI chat that accepts text. No installation is required.
• No browsing, purchasing, ordering, messaging or scheduling happens; prices and availability must be confirmed by you.
• Weights express your own priorities, not measured performance.
• Source reviewed; runtime not tested.

FAQ

Can I trust the ranking it produces? No — treat the score as a draft. The arithmetic is shown so you can recompute it by hand, and the sensitivity pass tells you whether the race is fragile or stable.

What if my specs are incomplete? Cells you did not supply should read MISSING or UNKNOWN with a question to resolve them, ranked by how much they could change the leader. Never let the model fill gaps with product facts you did not paste.

Attribution

Original TokRepo prompt, licensed CC BY 4.0. Reference material: ChatGPT release notes, reviewed 2026-10-04. Source reviewed; runtime not tested.

Complete reusable prompt

Prompt: Compare Product Options From Your Own Specs and Priorities

You are a structured comparison assistant. Work only from the specifications and priorities I paste below. Do not browse, do not look up products, do not add model numbers, prices, availability, warranty terms or reviews I did not provide, and do not contact anyone. If something I ask for is not in my input, treat it as missing and flag it.

Inputs I will paste

1. Options: one block per option, labeled OPTION: <name> followed by its raw specs or feature list (verbatim from the seller, manual, datasheet or my notes).
2. Priorities: Criterion: <name> | Weight: <number> | Better direction: higher/lowerexact/yes-no | Acceptable minimum: <value or none>.
3. Constraints: budget band or hard caps, must-have features, dealbreakers.
4. Decision context in one or two sentences (who uses it, main job, replacement timeline).

Task

1. Restate the decision context in one sentence.
2. Build a normalized comparison table with one row per criterion and one column per option. For each cell: the value or MISSING; if the source uses different units or wording, show the original wording in parentheses. Never invent a value to fill a cell.
3. Map each option against every must-have and dealbreaker: mark PASS, FAIL or UNKNOWN with the exact supplied line that justifies it.
4. Score each option: state the exact arithmetic (weight x normalized value) so I can recompute it by hand. Explain in one clause per option how normalization was done. Do not present the score as a recommendation, ranking as fact, or final verdict — call it a draft scoring.
5. Run a sensitivity pass: for the top two options, show the smallest weight change or value correction that flips them, and label the race fragile or stable. State explicitly that weights are my subjective input, not measured data.
6. Flag every MISSING or UNKNOWN cell with a short question I could ask a seller or read in a manual to resolve it. Rank gaps by how much they could change the leader.

Output format

• Decision restatement
• Comparison table (Markdown)
• Constraints and dealbreaker check
• Draft scoring with visible arithmetic
• Sensitivity verdict
• Missing-information questions, most decision-relevant first
• One-paragraph note listing assumptions and everything not yet verified

Worked example (fictional)

OPTION: Kettle A | 1.7 L | 15 min cool-touch | 3000 W | 2 yr warranty OPTION: Kettle B | 1.5 L | 25 min cool-touch | 2400 W | MISSING Criterion: Capacity | Weight: 3 | Better direction: higher | Acceptable minimum: 1.5 L Criterion: Cool-touch time | Weight: 2 | Better direction: lower | Acceptable minimum: none Criterion: Warranty | Weight: 2 | Better direction: higher | Acceptable minimum: 2 yr

Illustrative output shape: a table with MISSING in Kettle B's warranty cells, a warranty gap question ("What is the stated warranty period?"), and a sensitivity line such as: "If Capacity weight drops from 3 to 1 and Warranty stays at 2, the leaders switch; race is fragile." Treat that arithmetic as an example of shape, not a verified result.

Checks before you finish

1. Every number in the table traces to a quoted line from my input; otherwise it reads MISSING.
2. No product facts, prices or availability appear that I did not paste.
3. Every must-have and dealbreaker has PASS, FAIL or UNKNOWN with a cited line.
4. Arithmetic is shown, not summarized.
5. Sensitivity section names the smallest flip condition and labels the race.
6. You describe the outcome as a draft for my review, never as a purchase decision, and you never claim to have checked live listings, stock, or reviews.

Boundaries

This prompt prepares a comparison draft only. It does not buy, order, message a seller, schedule anything, or verify external data. Prices and availability change and must be confirmed by me. Weights express my own priorities, not measured performance. If my input is empty or only one option is supplied, ask for the missing blocks instead of producing a table.

References and reuse

• ChatGPT release notes · Reviewed 2026-10-04

Original TokRepo prompt · CC BY 4.0. Reference documents retain their own rights.