| name | addictive-blueprint |
| description | Score a proposed Pixel Hoops gameplay feature against the addictive-gameplay blueprint. Use when the user asks whether a feature, mechanic, screen, or change "follows the blueprint", "is addictive / engaging / sticky", "earns its place", or wants a design review of a gameplay idea against the roguelike addiction principles. Reads docs/addictive-blueprint.md as the source of truth. |
Addictive Blueprint Review
Score a proposed gameplay feature against Pixel Hoops' addictive-gameplay blueprint and report where it passes, where it falls short, and how to fix the gaps.
This skill is a design tool, not a rubber stamp. Be honest and specific. A feature that scores poorly should be told so, with concrete fixes, not flattered.
Step 1: Load the source of truth
Read docs/addictive-blueprint.md. The checklist near the bottom (18 criteria as of the collection-pacing amendment) is authoritative; if it has changed, use the changed version, not the list reproduced anywhere else, and take the passing bar from the checklist's own intro line. Also skim docs/gameplay-redesign.md for how the current game works (auto-sim 5-on-5, agency in roster, game plan, and the branching run map), so scoring is grounded in the real game, not a generic one.
Step 2: Identify the feature under review
The feature can come from any of these, in priority order:
- Text the user provided directly (in their message or as the skill argument).
- A document or section they point to (read it).
- The current working changes, if they ask to review "this" or "the diff": run
git diff (and git diff --staged) and infer the feature from the changes.
If the feature is unclear or underspecified, ask one short clarifying question before scoring. Do not invent a feature.
Step 3: Score each criterion
For each criterion in the checklist, assign:
- pass: the feature clearly satisfies the yes/no test.
- partial: it satisfies the spirit but has a gap or depends on something not yet designed.
- fail: it does not satisfy the test.
- n/a: the criterion genuinely does not apply (use sparingly, and justify it). An n/a does not count against the feature but also does not count toward the passing bar; say so.
Judge against the real game. For example, "immediate feedback" should reference the juice systems the game actually has (count-up scores, screen shake, callouts, haptics). For each criterion give a one-sentence justification tied to the feature, not a restatement of the criterion.
Step 4: Report
Output, in this order:
- Verdict line: the feature name, the score as
X / N passed against the checklist's current criterion count (count pass only; note partials and n/a separately), and whether it clears the bar the checklist's intro line states (14 of 18 as of the collection-pacing amendment).
- Scorecard table: one row per criterion: number, short name, verdict, one-sentence justification.
- Top gaps: the 2 to 4 most important
fail or partial items, each with a concrete, specific fix phrased as an action ("add a count-up on the recruit's stat reveal", not "improve feedback").
- Strengths: 1 to 3 things the feature does well, so they are preserved.
- One-line recommendation: ship as-is, ship with the listed fixes, or rework.
Keep the report tight and skimmable. Favor specific, buildable suggestions over general design advice.
Example scorecard row
| # | Criterion | Verdict | Why |
|---|
| 2 | Immediate feedback | partial | The recruit screen shows the new player but the stat reveal does not pop, count up, or fire a haptic, so the reward does not land. |
Notes
- The checklist is the contract. If the user wants to add or change a criterion, edit
docs/addictive-blueprint.md (the source of truth), not this skill.
- Implementation-level feel review lives in the
feel-review skill (source of truth: docs/feel-conventions.md); this skill stays at the design level, so when a request is about a built change's feel rather than a feature's design, use feel-review instead.