Mavera DocsThe prompt library
Copy-paste prompts for the fifty most useful jobs, organized by what you're trying to do.
Turn a prompt into a real run
Choose an audience, edit variables, run the prompt, and refine the answer.
Choose Mave from the sidebar.
Good afternoon
Choose a surface from the sidebar to begin.
- 1Open MaveChoose Mave from the sidebar.
- 2Add an audienceSelect a named audience so the prompt has a clear point of view.
- 3Edit the promptWrite or paste a prompt and replace its bracket variables.
- 4Run itSend the local prompt.
- 5Refine the answerAsk one sharper follow-up instead of starting over.
Steal freely. Replace anything in [brackets]. All of these work typed straight into Mave's chat.
Understand a market
Build an audience of [segment] and profile them: what they buy today, what frustrates them, what would make them switch.
What are the distinct customer types for [product category]? Build the two most valuable as audiences.
Run a focus group with [audience] on: "What's actually broken about how you [job-to-be-done] today?"
Test messaging
A/B test with [audience], 100 respondents: "[message A]" vs "[message B]". Winner, margin, and the top reason per side.
Show [audience] our homepage headline for a 5-second read: what do they think we sell? Percent who got it right?
Take our value prop — "[prop]" — and rewrite it in the customers' own language. Then test yours-vs-theirs.
Price things
Where does price resistance kick in for [product] with [audience]? Test [$X / $Y / $Z] between-subjects, 150 respondents.
Same price, three framings: [monthly] vs [annual-discounted] vs [usage-based]. Which triggers least resistance with [audience]?
What would we have to add to [product] for [audience] to accept a [20%] price increase without churn risk?
Work with your files
(attach) Extract every business fact from this deck and save the good ones to context.
(attach) Summarize this 40-page report in one page for a busy executive. Keep every number that matters.
(attach two files) Diff these proposals: what changed, and which version is stronger for a risk-averse buyer?
(attach CSV) Three most surprising things in this data. Chart the one that matters most.
(on Ingest) This is a completed customer study. Extract a source-quoted dossier. Separate stated claims from inferred readings and do not create a new audience automatically.
Produce deliverables
Turn this conversation's findings into a client-ready report: summary, method, findings with numbers, quotes, recommendations.
Make that report a PowerPoint — one finding per slide, our brand colors, no clip art energy.
Generate a hero image for [use]: [description]. Use our brand palette.
Write the launch email for [thing] in our brand voice. Then write the version for skeptics who ignored the first one.
Delegate research
(Workflows page) Figure out whether [audience] would pay for [product] and which feature to lead marketing with.
(Programs) Build a GTM blueprint for [product]: market map, competitors with real data, positioning, pricing strategy, launch sequencing. Cited sources.
(Skills page) Save this as a skill called "[name]": [your repeatable procedure, written like you'd brief a person].
Keep yourself honest
Play devil's advocate on everything we just concluded. Strongest three counter-arguments, steelmanned.
What in this analysis is grounded data vs your synthesis? Label each finding.
What real-world signal would prove this study result wrong? What should we watch?
Is the difference between [A] and [B] meaningful at this sample size, or noise? Answer statistically.
Manage the account
What did we spend this week, on what, and is anything trending up unusually?
Set a monthly limit of [N] credits on this project.
Share a read-only link to this conversation I can send to a client.
The meta-prompt
When you don't know how to ask: "Here's what I'm actually trying to decide: [decision]. Ask me what you need to know, then propose the cheapest research plan that would genuinely change my mind one way or the other." Fable 5 is good at this — the reasoning upgrade means it designs research, not just runs it.