For PokemonAutomation/ComputerControl#259

Goals

• Run E2E reliably on ENG version; should work with JPN version in theory, with necessary support for patches
• Milestones:
  • Regular, minimal playthrough (upto Mewtwo)
  • With optional game content, catches, etc
  • With built-in integration with RNG programs

Requirements

• Reliable: should one-shot without retries most of the time; retry without exiting by default
• Configurable: to extent to different mode, options, integrations with RNG programs
• Recoverable / Resumable: for testing as well as for unexpected incidence
• Testable: give us confident and make it easier to maintain
• Modular: provide building blocks that can benefit other autostory / FRLG programs

See https://github.com/dayfine/Arduino-Source/blob/main/docs/design/AutoStory-Architecture.md for more details

Design Principles

• rely on pret/pokemonfirered for in game data to make reliable decisions
• handle random events well
• different layers of abstraction:
  • shared autostory infra
  • shared 2D game nav infra / map serving infra
  • per-game map and moving logic
  • per-game detectors
  • per-game actions (how to do X)
  • per-game story events / table

Size

~50 PRs and ~20k lines, about half of which has been done:

See https://github.com/dayfine/Arduino-Source/blob/main/docs/design/AutoStory-FRLG-Sizing.md for more details

References:

• Design: AutoStory-Architecture.md
• Navigation layers in AutoStory-Navigation-Stack.md.

MVP PRs already coded: to be upstreamed

• Shared navigation core → shared story steps
• FRLG maps and planner → walker: encounters and obstacles → walker: position recovery → AutoStory: new game to the Boulder Badge

Milestones (not yet coded) are tracked in a fork issue:

• 1st Gym to Hall of Fame (M8)
• Run with options (M10)
• Sevii Islands and Mewtwo (M11)
• RNG integration (M12).