Typing and intents

How commands, modal fields, and event payloads use TypeScript, and where intent-aware typing is already strong or still aspirational.

Strong areas today

Chameleon already has good type inference in a few places:

• event payload narrowing by event name and event.type
• command option inference through command(...), subcommand(...), and opt.*
• modal field inference through modal(...).add(...).execute(...)
• typed component contexts for many interaction flows

Slash command options

const ping = command('ping', 'Ping a user')
  .user('target', 'User to ping', { required: true })
  .boolean('loud', 'Whether to ping loudly')
  .execute(async (ctx) => {
    const user = ctx.options.target
    const loud = ctx.options.loud

    await ctx.reply({
      content: loud ? `PING ${user.id}` : `ping ${user.id}`,
      ephemeral: true
    })
  })

ctx.options.target is inferred from the option type, not manually cast.

The newer command DSL also preserves richer inference for:

• mentionable as User | Role
• attachment as an uploaded Attachment
• selected values from choices(...)
• nested subcommands and subcommand groups

Modal fields

const profileModal = modal('profile', 'Profile')
  .add(
    field.short('name', 'Name'),
    field.checkbox('accept_rules', 'Accept rules'),
    field.fileUpload('attachment', 'Attachment', { required: false })
  )
  .execute(async (ctx) => {
    ctx.fields.name
    ctx.fields.accept_rules
    ctx.attachments.attachment
  })

Intents: design goal vs current reality

The roadmap wants compile-time intent awareness. Conceptually that means using missing intents as a type error, not discovering the problem only at runtime.

The current codebase is directionally aligned with that goal, but it is not fully realized across the entire framework surface yet. Treat this as a design target rather than a solved feature everywhere.