# Crafting

The core gameplay loop. Where the player's hands are most of the time.

## Stations

Workshops unlock as you hire matching Engineer specialties. Working list:
- Forge
- Alchemy Table
- Loom
- Enchanting Circle
- Kitchen
- Glassworks
- Jeweler's Bench
- Tannery
- Carpenter's Shop

Each station has tiers (I → V) that you upgrade with materials + a master Engineer. Higher tiers unlock more slots, higher-quality recipes, faster crafts.

## Mechanic

Drag ingredients into station slots, hit craft.
- Each station has 3–6 ingredient slots depending on the recipe.
- A finished item typically needs 4–5 components.
- Components are themselves crafted from upstream raw materials. Multi-tier supply chains — see [14-materials.md](14-materials.md) for the materials taxonomy and Raw → Refined → Component → Final pipeline.

After Craft is hit, the result auto-resolves on the craft's timer. **Master's Touch** ([16-engagement-layers.md](16-engagement-layers.md)) is an optional skill-timing mini-game that runs during the resolution window — two-meter input that nudges quality and crit chance up. Skippable; veteran auto-craft is a first-class workflow.

### Example: Moonsilver Dagger
Final craft at the Forge:
- Refined Moonsilver Ingot (Forge — from raw moonsilver ore + flux)
- Tempered Hilt (Carpenter's Shop — from seasoned hardwood + bone)
- Wrapped Grip (Tannery — from cured leather + waxed thread)
- Balance Jewel (Jeweler's Bench — from cut gem + silver wire)

Each component is its own craft. Each upstream material came from gathering or earlier crafts. The player can outsource components to engineer minions running queues, or make them by hand for higher quality.

## Quality bands

Crude → Common → Fine → Masterwork → Legendary

Output quality is a function of:
- Average ingredient quality
- Station tier
- Assigned engineer's skill + subspec match
- Trait modifiers (the Lucky engineer; the Drunk one)
- Luck roll

Critical successes promote one band. Critical fails demote. Quality drives sale price, equipment effectiveness, and patron contract eligibility.

## Discovery

Three ways to learn recipes:
1. **Library research** — Scholars convert time + reagents into recipe unlocks.
2. **Patron / expedition rewards** — recipes drop as loot.
3. **Experimentation** — drag off-recipe ingredients into a station and see what happens.

Experimentation outcomes:
- Most produce generic byproducts ("Failed Decoction").
- Some accidentally hit a real catalog recipe → **"New Discovery!"** flourish, recipe added to your book.
- Scholars generate **hints** ("Theora believes moonpetal + ghoul-fang at the alchemy table may yield a sleep draught") — these constrain experimentation enough to keep it fun without removing the joy.

## Catalog

The game ships with a fixed catalog of craftable item *recipes* (~500–1000 target). Players don't know all recipes; experimentation reveals which combos map to which catalog entries.

Crucially, **items are composed visually at runtime from layered sprites** — see [07-item-cards.md](07-item-cards.md). The same recipe instantiated with different component choices, materials, and quality outcomes produces visibly different cards. Two players who craft the same falchion recipe but use different pommel gems get visibly different swords. This means "uniques" are the norm, not the exception.

## Open questions

- How do we surface the recipe book without it becoming a spreadsheet? Maybe by-station view, by-discipline view, "in progress" filter?
- How aggressive is the experimentation penalty? Wasting rare ingredients should sting a little but not feel cruel.
- Quality at finished-item level: does it cascade fully from components, or can a great Engineer save a Crude component? Probably some of both.