enemy-kit-balance · diff
git:20260912.28c7171 to git:20260913.74cd25b
18 added, 35 removed. Audit A to A.
---
name: enemy-kit-balance
- description: Each foe is a small moveset with tells, punish windows, and cooldowns. Ask poke-guard vs burst-duelist vs swarm-chaff vs artillery. Do not balance by shrinking player iframes or growing silent hitstun.
+ description: Use when an enemy moveset has unreadable pressure, weak punish opportunities, repetitive loops, excessive overlap with nearby enemies, or balance changes are leaking into unrelated player timing systems.
---
- # Enemy kit balance
-
- Ask the column:
-
- | Column | Kit shape |
- |---|---|
- | poke-guard | short hits, long idle |
- | burst-duelist | one big commit, long recover |
- | swarm-chaff | weak, many, no armor |
- | artillery | safe at range, weak up close |
-
- ## Budget (write these four numbers per move)
+ # Enemy Kit Balance
- 1. Telegraph / startup — must be readable.
- 2. Active — box on.
- 3. Recovery — punish window.
- 4. Cooldown / stamina — how soon it repeats.
+ This skill owns **individual enemy move budgets and kit trade-offs**. `attack-tell` owns warning channels, `group-tactics` owns multi-enemy concurrency, and `difficulty-design` owns difficulty scaling policy.
- Plus: damage, hitstun column, armor or none, range.
+ ## Modes
- A kit gets **one** scary move. The rest are pokes or movement. Two unreactable heavies on one body is a bug unless the column is a raid-clock boss.
+ | Mode | Kit emphasis |
+ |---|---|
+ | poke-guard | frequent low-commitment pressure with authored gaps |
+ | burst-duelist | higher commitment burst and punish windows |
+ | swarm-chaff | simple low-cost moves intended for groups |
+ | artillery | range pressure with authored close-range weaknesses |
- ## How to tune instead of cheating
+ ## Move budget
- | Problem | Do | Do not |
- |---|---|---|
- | player dies too fast | cut damage or overlap | grow player hitstun |
- | player never punished | lengthen recover or shorten armor | delete dodge-iframe |
- | move unreadable | grow startup / add pose+sfx tell | add more HP |
- | same move loops | add cooldown or a whiff punish | shrink input buffer |
- | crowd is unfair | cap alive (spawn-wave) | give every chaff a grab |
+ For every move publish logical startup/tell handoff, active interval, recovery, repeat/cooldown resource, damage/reaction, range and any armor/guard interaction. Values are project data; clip length does not substitute for the logical table.
- HP and damage follow difficulty-design. Feel columns stay still. See balance-design rule 5.
+ A kit may have one or several high-impact moves depending on encounter role. What matters is that their combined frequency, overlap, coverage and counterplay fit the encounter budget. Do not impose a universal “one scary move” law.
- ## Tells
+ Punishability is defined by the selected game's legal response windows: movement escape, block/parry, interrupt, reposition, resource response, single hit, combo, and so on. Do not require every whiffed heavy to permit a universal two-hit confirm.
- Startup must show on pose, audio, or a published wind-up. Juice-only red flash is not a tell if a11y flash-off is on.
+ ## Encounter interaction
- Player punish: walk-in or a published light that fits inside recovery. If the player cannot complete a 2-hit confirm on a whiffed heavy, the recover is too short for that column.
+ When multiple enemies can pressure simultaneously, combine the kit table with `group-tactics` concurrency limits. A move that is fair in a duel may be invalid when several copies overlap. Tune the enemy or encounter budget before secretly shrinking player dodge, input buffer, hitstop or recovery contracts.
- ## Accept
+ ## Acceptance
- - Designer can name the scary move and its punish
- - Two adjacent foes prefer different player answers (dodge vs poke vs close the gap)
- - A patch note changes kit numbers, not the player's clock
+ Record move-table data and the intended counter for each high-impact action. Test the kit alone and in representative encounter compositions from `group-tactics`. Measure warning-to-activation, punish windows, repeat cadence and overlapping unavoidable threat. A patch should be explainable as kit/encounter data changes rather than hidden edits to unrelated player timing.