MiroFish Tutorial for Scenario Simulation

Best-fit use cases

• A product team wants to model launch objections before publishing a roadmap.
• A research team wants to compare trading narratives without building a full platform.
• An open-source user wants to self-host but still needs a checklist for graph and memory services.

Operating steps

1. Start with the product launch, trading, public reaction, or prediction market template.
2. Replace sample actors with the people, groups, or market participants in your case.
3. Pick hosted checkout for speed, GitHub self-host for control, or offline notes for private planning.
4. Run the minimum useful simulation first, then increase agents or rounds only after the report is readable.
5. Capture follow-up prompts so future runs can compare the same scenario over time.

Common risks

• Tutorials that skip deployment choices can leave users with a good prompt but no working workspace.
• Self-hosting without memory and graph checks can produce empty or repetitive results.
• Public reaction simulations should be framed as decision support, not certainty.