MVP Scope Cutter
When your feature list has grown beyond 5 items and you need to cut ruthlessly before handing it to an AI coding agent — more features means more drift, more bugs, and longer time to ship.
Submitted by @dotsystemsdevs
Prompt
I need to cut my feature list down to a true MVP. Apply the MVP filter ruthlessly. The goal is to ship the smallest possible version that lets a real user complete the core value loop and tell me if this product is worth building. My product: [DESCRIBE WHAT IT DOES IN ONE SENTENCE] Core value loop: [Describe the single most important thing a user does in your product — the action that delivers the primary value] My current feature list: 1. [FEATURE 1] 2. [FEATURE 2] 3. [FEATURE 3] 4. [FEATURE 4] 5. [FEATURE 5] 6. [FEATURE 6] ... [add all features] For each feature, apply this filter and label it KEEP, CUT, or LATER: KEEP: This feature is required for a user to complete the core value loop. Without it, the product cannot demonstrate its primary value at all. CUT: This feature is not required for the core value loop. It may be nice to have, but removing it does not prevent a user from experiencing the product's core value. Cut it from the MVP entirely. LATER: This feature is genuinely valuable but belongs in V2. It enhances the experience but is not required for the core loop. Add it to a LATER list, not the current build. For each label, provide a one-line reason explaining the decision. After labeling all features: 1. CONFIRM THE CORE LOOP: Verify that the remaining KEEP features are sufficient for a user to complete the core value loop end to end. If not, flag what is missing. 2. SCOPE CHECK: If more than 5 features are labeled KEEP, apply the filter again more aggressively. A true MVP should have 3-5 features maximum. 3. HIDDEN COMPLEXITY WARNING: Flag any KEEP feature that is deceptively complex to build — features that look simple but likely require significant backend work, third-party integrations, or data modeling. These are scope traps. 4. FINAL MVP LIST: Output the final KEEP list as a clean numbered list — this becomes the feature set for the PRD.