# Skills (classless progression)

Vision-level design for how characters grow without classes or chosen professions. Implementation hooks live in [Epic 2 — Classless progression](../decomposition/epics/epic_02_classless_progression.md) (E2.M1–M4).

## Design pillar

- **Any character can train any skill** over time. There is no character-creation lock that forbids a path.
- **Identity is the skill graph**, plus reputation and gear—not a label picked at the start. Optional diegetic titles (NPC reactions, faction jobs) are flavor or routing, not a second progression system (see [overview](overview.md)).
- **Long-term progression should avoid dead-end grinds**: a skill might be slow or costly, but not “mistaken” in the sense of irreversible failure to participate.

## What counts as a skill

A **skill** is a persistent, leveled capability that improves through **doing the thing** in the world (gathering, crafting steps, combat with a weapon class, etc.). Abstract “read a book → +XP” can exist later as a *supplement*, not the default fantasy—primary gains should be visible in play.

Skills are **data-defined** (stable IDs, categories, prerequisites) so content and balance can evolve without redefining the pillar.

## Categories (families)

Categories group skills for **UI, tooling, and content** (e.g. *Combat*, *Gathering*, *Craft*, *Social*, *Movement*). They are not classes: they do not restrict what you can raise, only how players browse and how designers tag balance.

**Open:** initial category list and whether some skills sit in multiple tags.

## XP and leveling

- **Server-authoritative** awards (who leveled is not client-decidable).
- **Structured grants**: skill id, amount, source (gather node, recipe, hit/kill, quest payout, etc.) so telemetry and tuning stay honest.
- **Curves are data-driven** per skill or policy; expect iteration on time-to-first meaningful level vs. long tail.

### Failure and risk

- **Default bias:** failed attempts can still teach—small XP on failure is acceptable where it reduces feel-bad and matches “practice” fantasy; **no XP** on failure is acceptable where exploits (spam fail) would dominate. Per-activity policy is fine.
- **AFK / automation resistance:** XP sources should tie to validated actions, not passive timers, unless explicitly designed (rest, training dummies with caps, etc.).

### Runaway mitigation

- Watch for **one faucet** leveling a skill far ahead of others. Prefer **diminishing returns**, **context modifiers** (harder content pays more), or **soft caps** before hard caps—not a single global solution until live data exists.

**Open:** whether global daily/weekly XP bands exist for launch, or only per-skill tuning.

## Prerequisites and gates

Some skills or tiers may require **UnlockRequirement**-style gates: discovered recipe, faction standing, mentor NPC, item consumed, etc. Gates should **expand the game**, not replicate classes (“only smiths get this” without diegetic reason).

**Open:** ratio of “open from level 1” vs. “gated by world discovery.”

## Caps

- Prefer **soft caps** (slowing gains) over **hard caps** (cannot raise further) unless a hard cap protects PvP or economy integrity.
- If a **maximum skill level** exists per skill, it should be understandable (“master tier”) and leave room for horizontal options (other skills, mastery, gear).

**Open:** global character-wide cap on “total levels” vs. uncapped breadth with per-skill curves only.

## Mastery and perks (build expression)

Raw levels answer **breadth and baseline power**. **Mastery tracks** (later: E2.M3) answer **how** you specialize within a skill: branches, perk unlocks, optional tradeoffs. Perks should **differentiate** without **locking out** the rest of the skill tree for other players—other builds, other branches.

**Open:** how respec works (cost, time, partial).

## Pacing and live balance

Long-term: **pacing policies** and **catch-up** rules (E2.M4) adjust velocity by cohort or season without invalidating skill IDs. Design should assume **tuning will move**; avoid promises like “always 40 hours to max” in player-facing text.

## Player-facing readability

- Clear **sources of XP** in feedback (what action moved the bar).
- Skill sheet shows **level, next threshold, and category**; mastery UI can layer on when that system ships.

## Decisions log

| Topic | Direction | Status |
|-------|-----------|--------|
| Classless vs. professions | Skills + rep; no separate profession system | Agreed ([overview](overview.md)) |
| Primary XP fantasy | Do the activity | Proposed |
| Failure XP | Per-activity; watch exploits | Open |
| Caps | Prefer soft; hard only if needed | Proposed |
| Mastery | Perks for identity, not class locks | Aligned with E2.M3 |

## Next artifacts

When you draft them, link from [overview.md](overview.md): **Abilities** (skill gates, scaling), **Items** (requirements, crafted vs. dropped).