How to use this guide
Read the beginner guide first if the opening run still collapses before Act 2. This page assumes you already understand the run loop, then focuses on build identity, progression choices, relic evaluation, and mode-specific pressure for Demo 0.5.1.
Build engines, not fixed rosters
The strongest teams are not collections of individually strong heroes. A coherent composition produces a trigger repeatedly, converts it into value, and survives long enough for that conversion to matter.
| Part | Question | Warning sign |
|---|---|---|
| Trigger | What happens several times each fight? | Only one hero can start it |
| Payoff | What value does each trigger create? | The build applies effects but does not win |
| Time | How does the team survive until the payoff scales? | The engine starts after the frontline dies |
| Access | Can the right hero reach the right target? | Damage is spent on a low-priority body |
Practical frameworks
- Shield and Stall. Durable first contact, repeatable shielding, delayed payoffs.
- Crit sustain. Fast Duelist or Assassin attacks convert Crit into healing, mana, or durability when the right relics appear.
- Status control. Poison, Frost, and Burn become strong when several effects reward the same debuff.
- Rush pressure. Needs immediate target access; a unit that spends the opening walking contributes nothing.
When to pivot
- Pivot when two or more owned pieces support a different repeatable trigger.
- Hold the plan when only one shop offer suggests the new theme.
- Drop a class tag that adds no payoff, protection, or target access.
- Change positioning before replacing a hero that is failing only because of its lane.
Hero progression: Rank C to S
Guildrun separates hero development inside a run, permanent unlocks across runs, and difficulty progression. Keep those layers distinct so you know what resets and what carries forward.
| Rank | Decision | Pick test |
|---|---|---|
| C | Place the base hero and confirm its job | Can this hero perform its job with the current formation? |
| B | Choose one of three fixed specializations; some add a second class | Does the path solve today’s fight or unlock an owned interaction? |
| A | Draft the first class modifier from active classes | Will this modifier trigger often enough to matter? |
| S | Take a second modifier and an extra item slot at Demo max | Is finishing this hero stronger than the next board-size breakpoint? |
Do not spread ranks evenly by default. A first Rank B on the main carry or anchor often changes more than giving a weak fourth hero a small bump. Dual-class paths expand later modifier pools, which increases flexibility but can make specific offers less consistent.
Grow the board from three to five active heroes only when the new slot has a job. The roster can hold six for a reserve specialist or Backup interaction. Team Size costs 45 Shards—compare that with the duplicate, item, or relic that would strengthen the current core immediately.
Respec with evidence
- Name the current hero’s job.
- Identify which part of the specialization no longer activates.
- Confirm the replacement path works with owned items and relics.
- Reposition before spending on the rebuild.
How to evaluate every relic offer
A relic should be judged by activation frequency, value per trigger, and the probability that the current roster can use it. Relics cannot be sold, so a narrow legendary can be more expensive than its Shard price suggests.
| Question | Strong answer | Weak answer |
|---|---|---|
| Who triggers it? | Two or more active heroes | A future recruit you do not own |
| When does it pay? | During the next important fight | After several unlikely conditions |
| What does it solve? | A named survival or damage failure | It is simply a higher rarity |
| What does it cost? | Fits the next Shard breakpoint | Consumes the budget for a needed rank |
Separate stability from ceiling. Defensive relics can protect a fragile run immediately. Economy and scaling relics may produce a higher ceiling only if the team survives the delay. Quest relics need a realistic completion route: count how often the current board creates the trigger in an ordinary fight before buying the quest.
Exact numbers can move between patches. Re-read the in-game text after a balance update, then filter the relic database or start from a status page to find matching pieces.
Red Rift: mission pressure and safer pivots
Red Rift combines stronger enemies with extra mission pressure. A team that eventually scales in normal play may fail because it loses too many heroes on the way to that ceiling.
- Value sustain earlier, but prefer sustain that also advances the engine.
- Treat hero deaths as a resource: a clear can still be strategically poor if it burns the mission allowance.
- Build the first reliable checkpoint before adding a second fragile carry.
- Avoid liability units that inflate deaths without changing the fight.
| Need | Prefer | Avoid relying on |
|---|---|---|
| Opening stability | Immediate shields, control, or durable first contact | A quest that completes after several risky fights |
| Damage | Repeatable triggers with target access | A carry that spends Rush walking |
| Recovery | A formation change with a measurable result | Repeating the same board through Rewind |
| Scaling | Growth that also helps the next checkpoint | Endless-only ceiling before stabilization |
Endless mode: build for repeatable scaling
The normal route asks whether a build can clear the next encounter. Endless asks whether the engine keeps producing more value as enemy health and damage continue to rise.
| Layer | Endless question | Failure signal |
|---|---|---|
| Growth | What increases again after another fight? | Every bonus is already fixed |
| Survival | How does the team handle larger hits and the anti-stall clock? | The frontline only delays defeat |
| Access | Can damage still reach priority enemies as formations change? | The carry is trapped walking or hitting summons |
| Redundancy | What happens when the main carry is controlled or blocked? | The entire engine stops with one hero |
Separate flat power from compounding power. One-time stats can bridge early waves; deep runs need repeatable triggers, permanent-in-run gains, or percent-based defenses. Storm begins at 50 seconds and escalates; Riftbreaker appears at 90 seconds—indefinite survival is not a win condition.
Change one layer at a time
- Open the exact stage formation and enemy variants.
- Confirm first contact and target access.
- Identify whether the loss is damage, survival, control, or scaling.
- Replace the weakest engine link, not the highest-stat hero by default.
Connected tools
Use the beginner guide for first-run fundamentals, the growth route for board size and upgrade order, hero dossiers for Rank B paths, fight modes for challenge thresholds, and stage formations when a specific lineup breaks the plan.