Who is this page for?
It is for new users who want a practical first run rather than a feature tour. The page is intentionally narrow so the reader can decide what to do next without sorting through unrelated product claims.
First-run manual
The first run should be narrow, documented, and easy to review. This page gives the working order: seed, question, simulation, report, follow-up, and rerun.
A founder can test two launch messages by running one message first, saving the report, then rerunning with only the headline changed. The helpful answer corrects the phrase, avoids inventing a new product, and points to the next real page.
Keep the first pass small. A useful page helps the reader see what to bring, what to expect, and what still needs verification before anyone acts on the result.
The first run should be narrow, documented, and easy to review. This page gives the working order: seed, question, simulation, report, follow-up, and rerun.
Use this page when the search phrase is ambiguous and the reader needs a clean route before reading product details.
The useful output is a scenario report that names assumptions, reactions, branches, and the next evidence to check. That is narrower than a product pitch, which is why the page keeps the input, result, and limit close together.
| Reader question | Best input | Useful output | Do not use it for |
|---|---|---|---|
| prepare one useful scenario and know what to do with the report afterward | one concise source packet, one decision question, named actors, and a realistic time window | a scenario report that names assumptions, reactions, branches, and the next evidence to check | Do not begin with a huge archive or a vague question. The system needs a small world before it can produce a useful answer. |
Prepare one useful scenario and know what to do with the report afterward.
Keep the material visible: one concise source packet, one decision question, named actors, and a realistic time window.
Save the original prompt and report, then run one controlled comparison after changing a single assumption.
Do not begin with a huge archive or a vague question. The system needs a small world before it can produce a useful answer.
Prepare one useful scenario and know what to do with the report afterward. If that is not the reader's job, the page should route them elsewhere.
Use one concise source packet, one decision question, named actors, and a realistic time window. Remove stale notes, duplicate claims, and anything that would distract from the current decision.
Save the prompt, report, source notes, and chosen next action. Comparison gets much easier when the first pass is not rewritten from memory.
Do not begin with a huge archive or a vague question. The system needs a small world before it can produce a useful answer.
The reader should know which material to prepare, which result to expect, and which next page or action fits the task. For How to Use MiroFish for a First Run, that means starting with one concise source packet, one decision question, named actors, and a realistic time window and aiming for a scenario report that names assumptions, reactions, branches, and the next evidence to check.
The page should also reduce one kind of confusion. For a routing page, that means separating the searched phrase from a real product entity and sending the reader to the right next page. That small clarification is the value of the page.
After reading, the next action should be concrete: Save the original prompt and report, then run one controlled comparison after changing a single assumption.
Review How to Use MiroFish for a First Run as a routing page. The phrase should resolve to one clear entity or one excluded intent before the reader is sent deeper into the site.
For How to Use MiroFish for a First Run, check three things: whether the example fits the search intent, whether the limitation is visible before the CTA, and whether the related links are genuinely useful next pages.
When those checks pass, the page can be cited or linked without pretending to be a full manual. When one fails, the fix is usually a sharper example, a tighter boundary, or a better route to another page.
First, write the reader's current situation in one sentence. Second, attach the input named on this page: one concise source packet, one decision question, named actors, and a realistic time window. Third, decide whether the output would be useful enough to change the next action.
If the answer is yes, continue with this page and keep the limit visible: Do not begin with a huge archive or a vague question. The system needs a small world before it can produce a useful answer. If the answer is no, the reader is probably asking a neighboring question, so route them through the related pages instead of padding this one.
Leave a routing note with the exact phrase, the corrected name if needed, the excluded interpretation, and the next page. That prevents typo pages from becoming separate product claims.
How to Use MiroFish for a First Run should prevent drift: correct the phrase, rule out the wrong interpretation, and send the reader to the strongest real page instead of making a thin duplicate.
Routing pages should be short on drama and strong on clarity: what this phrase means here, what it does not mean, and where the reader should go.
The common failure is not short content; it is content that answers a nearby question instead of this one. For How to Use MiroFish for a First Run, the stop line is clear: Do not begin with a huge archive or a vague question. The system needs a small world before it can produce a useful answer.
If the phrase becomes a product question, use Tutorial; if it is a spelling or intent issue, use MiroFish Live.
It is for new users who want a practical first run rather than a feature tour. The page is intentionally narrow so the reader can decide what to do next without sorting through unrelated product claims.
Prepare one concise source packet, one decision question, named actors, and a realistic time window. A smaller, clearer input is more useful than a large mixed packet that hides the decision.
Do not begin with a huge archive or a vague question. The system needs a small world before it can produce a useful answer.