The Last Garden

What This Is

An idle game for the browser in which the player tends a walled garden at the fraying edge of a world that is being forgotten. Every plant grown recovers a memory fragment from a dissolving civilization, and the idle loop is the act of remembering. Tonally: Hayao Miyazaki directing a podcast about grief — warm watercolor visuals, a solo cello, and the slow-burn unfolding of a seven-Season story about loss, the cost of trying to erase suffering, and what it means to let something persist.

The audience is the lineage that responded to A Dark Room and Universal Paperclips — players who want the reason the numbers go up to be the thing they actually care about — overlapping with the cozy-gaming and narrative-game audiences who currently have no idle game built for them.

Core Value

Every idle mechanic must function as a metaphor that the player absorbs without being told. Composting is letting go. Prestige is grief. Roothold is what survives. If a mechanic does not carry narrative weight, it does not belong in this game.

This drives every tradeoff: when economy and meaning conflict, meaning wins. The numbers are in service of the story, not the other way around.

Requirements

Validated

(None yet — ship to validate)

Active

Core idle loop: planting, waiting, harvesting memory fragments
Offline progression — the garden grows while the player is away
Memory fragment system — every harvest yields a piece of authored narrative content
Seasonal prestige cycle — die-off resets surface garden, persists Roothold and learned memories
Seven authored Seasons (Soil → Roots → Canopy → Storm → Depth → Loom → Return), each with its own mechanic, story beats, and tonal shift
Three named characters (Lura, the Nameless Man, the Archivist) with the dialogue arcs the bible specifies
Cross-pollination, composting, ecosystem planting (Season 5+), and Memory Storms (Season 4+)
Place-memory vignettes from canopy trees (Season 3+) — short interactive scenes
Final binary narrative choice and "The garden persists" ending
Save persistence (local at minimum) so a 7-Season idle game can survive gaps in play
Watercolor-adjacent visual style and acoustic audio (solo cello main theme + ambient garden sounds + the high-tone silence of the Pale)
Color palette and audio that shift across the seven Seasons
Cosmetic monetization (planters, walls, gates, tool skins) — never gating narrative
Season acceleration purchases — never skipping story beats
Premium "Keeper's Journal" feature for annotating and organizing collected fragments

Out of Scope

Gacha mechanics — The game's thematic argument is that you cannot reduce complex things to simple transactions. Gacha would directly contradict the story.
Lootboxes — Same reason as above. The monetization model must not undermine the story's argument.
Narrative gating behind purchases — Story is the product. Story content is never paid.
Skipping (vs accelerating) Seasons — Players must never miss story beats; acceleration is allowed, skipping is not.
Generic fantasy flora — Plants must look like real-world species, slightly wrong. No D&D-style fictional flora.
Combat / boss fights — The Archivist is not a boss. There is no enemy.
Multiplayer / social features — This is a contemplative, solitary experience. Garden visiting / sharing may be considered post-1.0 if it does not break the tone.
Lore codex / encyclopedia entries — World-building emerges through fragments only. The player should always feel like they're almost understanding.
Voiced dialogue (v1) — Tone is "a friend texting you while you're at work." Text fits the medium. Reconsider only if the cello-and-silence soundscape benefits.
A Keeper with a name, backstory, or dialogue beyond the final choice — The Keeper is a presence, not a personality.
Generic cosmetic items unrelated to the garden setting — Cosmetics must reinforce, not dilute, the aesthetic.

Context

Market positioning. The idle game market is ~$14.2B (2025), projected to ~$34.8B by 2034. Idle RPGs are growing at 13.7% CAGR. Industry analysts identify narrative quality as the emerging differentiator. The genre clusters around three settings (generic fantasy combat, capitalism parody, sci-fi mining); the cozy-gaming-meets-narrative-idle space is essentially uncontested.

Lineage. A Dark Room (idle resource gathering wrapped around a slowly-unfolding mystery) and Universal Paperclips (an existential horror story disguised as a spreadsheet) are the proof points. Both became cultural phenomena with zero marketing. The Last Garden is consciously in this lineage and should be marketed within it.

Tonal anchors.

Visual: warm watercolor; hand-painted textures; plants as real species made slightly wrong; the Pale rendered as overexposed white silence; palette shifts by Season (golden/autumnal → deep green/storm → dawn/silver).
Audio: solo cello main theme; ambient garden sounds that thin as the Unremembering nears; near-silence and a faint high tone in the Pale.
Writing: short. Every line earns its place. No exposition dumps. Fragments do the world-building, and fragments are by nature incomplete.

Story structure. Seven prestige Seasons (Soil, Roots, Canopy, Storm, Depth, Loom, Return) with explicit mechanic + narrative beats per Season already mapped in the story bible. Roothold is the never-lost prestige currency representing what has been truly understood.

Pipeline reality. Solo / small-team development with AI-assisted art and audio production, hand-refined where it matters most. This means the production pipeline (asset generation → curation → integration) is a load-bearing piece of the system, not an afterthought.

Constraints

Platform: Web (browser) — Browser-based to match the A Dark Room / Paperclips precedent and remove install friction. Steam and mobile ports are post-1.0 considerations.
Engine/stack: Open — research will recommend (TypeScript-based 2D framework most likely; Phaser, PixiJS, or Godot's web export are leading candidates pending research).
Scope: Full 7 Seasons at v1 — committing to ship the entire authored arc. This is a multi-year scope and the roadmap will reflect it.
Art/audio production: AI-assisted then hand-refined — this constrains tooling choices (we need a curation/refinement pipeline) and asset pipeline design.
Save model: Local persistence required (idle game economics demand it); cloud sync via simple backend or Steam-cloud-equivalent is a stretch goal.
Monetization: Cosmetic-only, plus Season acceleration (not skipping), plus premium Keeper's Journal. No gacha, no lootboxes, no narrative gating. This is a hard thematic constraint, not a soft preference.
Audience: Cozy + narrative + idle Venn — the game must read as cozy at first glance and earn its emotional weight gradually. v1 must not lead with grief.

Key Decisions

Decision	Rationale	Outcome
Browser as the launch platform	Matches A Dark Room / Paperclips lineage; zero install friction; broadest reach for narrative-driven idle	— Pending
Full 7 Seasons in v1 (not vertical slice)	User explicitly chose to commit to the entire authored arc rather than ship Season 1 alone	— Pending
AI-assisted art + audio, hand-refined	Pragmatic for solo/small-team scope of 7 Seasons of authored content; refinement step is mandatory to preserve craft	— Pending
Engine deferred to research phase	No internal preference; let research surface the right fit for web + 2D + idle + content-heavy	— Pending
Single explicit narrative choice (final scene only)	Idle game players project themselves onto their systems; the garden choices are the characterization	✓ Good (locked by story bible)
Roothold as never-lost prestige currency	Roothold represents understanding, not progress; it's the one thing the Unremembering cannot touch	✓ Good (locked by story bible)
Cosmetic-only monetization with Keeper's Journal premium	The story is the product; monetization that gates narrative would undermine the game's thematic argument	✓ Good (locked by story bible)

Evolution

This document evolves at phase transitions and milestone boundaries.

After each phase transition (via /gsd-transition):

Requirements invalidated? → Move to Out of Scope with reason
Requirements validated? → Move to Validated with phase reference
New requirements emerged? → Add to Active
Decisions to log? → Add to Key Decisions
"What This Is" still accurate? → Update if drifted

After each milestone (via /gsd-complete-milestone):

Full review of all sections
Core Value check — still the right priority?
Audit Out of Scope — reasons still valid?
Update Context with current state

Last updated: 2026-05-08 after initialization

9.1 KiB Raw Blame History