atiran/sovereignx
3
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Composite flag docs (#5349)

Browse source
This commit is contained in:
Pawkkie 2024-09-09 04:43:15 -04:00 committed by GitHub
parent 2a27aad259
commit 32dde8af64
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 10 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
docs/ai_flags.md
View file

@ -27,6 +27,16 @@ If you are not using competitive syntax parties, instead access the trainer data
# What AI Flags does pokeemerald-expansion have?
This section lists all of expansion’s AI Flags and briefly describes the effect they have on the AI’s behaviour. In all cases, please check the corresponding function or surrounding code around their implementation for more details. Some of these functions are vanilla, some share a name with vanilla but have been modified to varying degrees, and some are completely new.
## Composite AI Flags
Expansion has two "composite" AI flags, `AI_FLAG_BASIC_TRAINER` and `AI_FLAG_SMART_TRAINER`. This means that these flags have no unique functionality themselves, and can instead be thought of as groups of other flags that are all enabled when this flag is enabled. The idea behind these flags is that if you don't care to manage the detailed behaviour of a particular trainer, you can use these as a baseline instead, and expansion will keep them updated for you.
`AI_FLAG_BASIC_TRAINER` is expansion's version of generic, normal AI behaviour. It includes `AI_FLAG_CHECK_BAD_MOVE` (don't use bad moves), `AI_FLAG_TRY_TO_FAINT` (faint the player where possible), and `AI_FLAG_CHECK_VIABILITY` (choose the most effective move to use in the current context). Trainers with this flag will still be smarter than they are in vanilla as there have been dramatic improvements made to move selection, but not incredibly so. Trainers with this flag should feel like normal trainers. In general we recommend these three flags be used in all cases, unless you specifically want a trainer who makes obvious mistakes in battle.
`AI_FLAG_SMART_TRAINER` is expansion's version of a "smart AI". It includes everything in `AI_FLAG_BASIC_TRAINER` along with `AI_FLAG_SMART_SWITCHING` (make smart decisions about when to switch), `AI_FLAG_SMART_MON_CHOICES` (make smart decisions about what mon to send in after a switch / KO), and `AI_FLAG_OMNISCIENT` (awareness of what moves, items, and abilities the player's mons have to better inform decisions). Expansion will keep this updated to represent the most objectively intelligent behaviour our flags are capable of producing.
Expansion has LOADS of flags, which will be covered in the rest of this guide. If you don't want to engage with detailed trainer AI tuning though, you can just use these two composite flags, and trust that expansion will keep their contents updated to always represent the most standard and the smartest behaviour we can.
## `AI_FLAG_CHECK_BAD_MOVE`
The AI will avoid using moves that are likely to fail in the current situation. This flag helps prevent the AI from making ineffective choices, such as using moves into immunities, into invulnerable states, or when the moves are otherwise hindered by abilities, terrain, or status conditions.