Pause when a Sephiria reward screen offers a third attractive damage effect. The choice is not simply which card is strongest. It is whether that card can fit into the active grid, receive useful Tablet levels, reinforce the weapon path already doing the work, and still leave the rabbit with a way to survive the next boss pattern.
This guide follows the full-release build and was checked on August 1, 2026, after the official 1.0.19 update and the 1.0.21 hotfix. The launch expanded Sephiria to six chapters, six weapon branches with more than 200 upgrades, roughly 300 Artifacts, roughly 70 Tablets, and more than 30 Potions. That scale is the reason this is a decision route rather than a permanent list of “best” drops.
Start 1.0 from town, not from an old mid-run save
The 1.0 notes state that an Early Access expedition saved in progress cannot continue in the official build; that run ends and the character returns to town. Do not treat that forced return as lost story progress or try to reproduce an obsolete inventory. Confirm that the title screen shows the current branch, then begin a fresh expedition from the town state the game preserved.
Open Lore → Tutorial before assuming an input or system disappeared. Version 1.0 added a reviewable Tutorial section, renamed some interface terms, added quest-location markers, and allows Tablets to be rotated with a middle-mouse click. Controller glyphs and font presentation also changed. If an older guide names a button that no longer matches the screen, the current tutorial and tooltip win.
Keep the first costume and Talent setup conservative unless the live cards clearly support a different experiment. Several costume effects changed in 1.0, so a 2025 recommendation can now describe the right costume name with the wrong tradeoff. Read both the benefit and penalty. If the penalty attacks the move you rely on—dash recovery, MP use, guarding, a special attack, or the damage channel named by the weapon—choose a neutral option for the learning run.
Give the build one working sentence
After the first weapon upgrade and Artifact choices, describe the build without item rankings:
“My weapon does damage through this action; these Artifacts reward that action; these Tablets raise the pieces that matter; this defensive tool lets me repeat it.”
A useful sentence might center on repeated normal attacks, a special attack, guarded retaliation, a spell, a follower, a debuff, or another effect explicitly named by the current cards. The exact engine changes every run. The test does not: each new reward must strengthen the trigger, strengthen the payoff, or buy the time needed to trigger it again.
Reject a reward when it creates a second unfinished engine. A powerful follower Artifact is not automatically useful when the inventory, Tablets, and weapon upgrades all reward direct weapon hits. An elemental card is not automatically a conversion if the tooltip never says the weapon or Artifact now uses that element. A special-attack bonus does not improve normal attacks merely because both come from the same weapon. Read the capitalized game terms literally.
Use the inventory’s Combo Effect panel as evidence, not decoration. If adding an Artifact advances the combo you chose and the next threshold is reachable without dismantling the core layout, it has a job. If the counter remains unchanged, the card may still be good, but it needs a separate reason to occupy an active cell.
Choose the next node by the build’s missing proof
The map is a sequence of tests, not a sightseeing route. Read the reward label on the live node and decide what must be proven before the boss:
- No engine yet: prefer a reward that can establish a weapon interaction, Artifact trigger, or Combo Effect. More currency does not repair a build with nothing worth buying around.
- Engine works but levels are scattered: look for Tablet or enhancement opportunities. The goal is to concentrate useful levels, not to fill every square with a different effect.
- Damage is adequate but mistakes are expensive: prioritize HP, defense, recovery, mobility, or a Potion opportunity confirmed by the current node and item text.
- The grid is clogged: route toward a merchant or conversion opportunity when one is actually shown. Carrying every mediocre Artifact forward makes later rewards harder to evaluate.
- The build is already boss-ready: money, a controlled optional fight, or a targeted upgrade can be justified. Do not take a harder encounter only because it appears to be the “advanced” choice.
Re-run the test after the node is cleared. If a reward changes the engine sentence, reorganize immediately. If it does not, do not let sunk cost decide the next three nodes. The first useful engine is a hypothesis; two supporting rewards make it a plan.
Treat every main-grid cell as active budget
Artifacts occupy the working inventory, while Tablets project positive or negative level changes into cells according to the pattern shown on their card. Placement therefore changes the value of an Artifact without changing the Artifact itself. Check four signals whenever the grid changes:
- Current level versus maximum level. A high-value Artifact already at its useful cap does not need another Tablet arrow.
- Direction. Rotate the Tablet and watch which cells change. Version 1.0 specifically changed the patterns or values of Competition, Wave, Preparation, and Daydream, so an Early Access screenshot is not a safe layout.
- Negative cells. Do not place the engine into a penalty because the overall grid looks tidy. A spare or utility item may tolerate that cell better, but its live card decides.
- Opportunity cost. A Tablet that gives small bonuses everywhere may be worse than one that pushes the core Artifact across a meaningful threshold.
Use a two-pass layout. First, place the two or three Artifacts that define damage and survival. Second, rotate and place Tablets around those anchors. Only then fill remaining cells. This prevents a common failure in which the inventory looks full while the pieces that determine boss performance sit at low levels.
Use the Side Bag as a quarantine zone
The Side Bag is a 1.0 feature unlocked through Destiny Inscription, and its contents are inactive. That makes it useful for uncertainty, not for hidden power. Move an item there when it could become valuable after one more reward, when it is being saved for conversion, or when it would disrupt the current layout while you compare options.
After moving an Artifact, check the Combo Effect counters and character statistics again. If damage, sustain, or a required combo fell, the item was part of the active engine even if you had mentally labelled it “spare.” Conversely, an inactive Side Bag item cannot justify a node route, weapon upgrade, or boss plan until it returns to the main grid.
Set an expiry rule: if the missing synergy has not appeared by the next major checkpoint, sell or convert the spare rather than protecting it for the entire expedition. The Side Bag should preserve a decision, not postpone every decision.
Make weapon upgrades pass an action test
Sephiria 1.0 has six weapon branches and more than 200 upgrades, and the launch patch changed numerous individual weapon effects. Do not compare upgrades by color or damage wording alone. For each choice, perform the action it modifies against ordinary enemies and answer three questions:
- What input or combat event triggers the effect?
- Which stat, element, Artifact, or Combo Effect supplies its payoff?
- Can the rabbit deliver it while still responding to the room’s attack warnings?
An upgrade fails the test when its damage is impressive only under a condition the build cannot maintain. A charge that repeatedly traps you inside telegraphed attacks, a guard effect on a route that has abandoned guarding, or an elemental conversion with no matching levels is not “bad”; it belongs to another run.
Do not generalize one weapon’s defensive input to all six branches. The 1.0.21 notes, for example, changed the Sword & Shield Cleave cost and made its guard frames active from the start of the animation. That is evidence about Cleave in the current Sword & Shield branch, not proof that every weapon can parry with the same timing.
Keep one dash for information you do not have yet
Rooms become dangerous when movement is spent before the attack pattern is understood. Enter at the edge, identify ranged enemies and persistent hazards, and let the visible warning determine the escape direction. A dash used to cross empty floor is no longer available when a projectile fan, ground marker, or rushing enemy removes the safe lane.
Separate positioning movement from emergency movement. Walk or attack from the side of the room that preserves open space. Spend a dash when it clears a confirmed hit, crosses a hazard with a known endpoint, or activates a build effect worth the risk. If the weapon upgrade asks for a dash, check whether another dash remains before repeating the loop.
At a boss door, stop moving long enough to audit the build:
- Can the main damage action hit while the boss is moving?
- Does the build have an answer when the safe window closes?
- Are Potions accessible rather than buried behind an inventory assumption?
- Did the last grid change lower the core Artifact or Combo Effect?
- Is a spare occupying active space while a useful item sits inactive?
If two answers are “no,” the build is not ready because of a specific missing job. Use that diagnosis on the next node or next run instead of searching for a universal S-tier item.
The Mystic Pot is a commitment decision
Version 1.0 changed the Mystic Pot. A single input rerolls into another Artifact while keeping its rarity; feeding the Pot two Artifacts produces one result from the next rarity. Older instructions that describe it as an Artifact-to-Tablet machine no longer apply.
Use the one-item conversion when an off-plan Artifact has no active job but its rarity is still worth preserving. Use the two-item conversion only when both inputs are expendable. “Not currently boosted” is not the same as expendable: remove each candidate from the grid, check the combo and statistics, and confirm that neither item supplies a hidden prerequisite before committing.
The Pot is not a rescue button for a confused build. A higher-rarity result can still demand the wrong trigger, element, or footprint. Convert to reduce dead inventory and create a new decision; do not promise that rarity will create synergy.
Spend the death on the first broken checkpoint
After a failed expedition, return to town and record the earliest point at which the engine sentence stopped being true. The final hit is usually too late to be useful. Use this checkpoint table:
| Observed failure | What to inspect | Next-run change |
|---|---|---|
| Rooms took too long before the build formed | First weapon trigger, first supporting Artifact, and whether early nodes chased currency without a purchase plan | Prioritize one engine proof before optional scaling. |
| The grid was full but damage remained flat | Core Artifact levels, Tablet direction, combo thresholds, and inactive Side Bag pieces | Anchor the core first and remove one unsupported effect. |
| Ordinary rooms were safe but the boss ended the run | Damage access while the boss moves, reserve dash, recovery window, and the upgrade’s required action | Replace one room-clear luxury with a boss-access or survival job. |
| A conversion made the build worse | The removed items’ combo counters and stat contributions | Perform the remove-and-check test before using the Mystic Pot. |
| A saved Preset loaded incorrectly | Client version and the applied inventory state | Update to at least 1.0.21, which fixed Save/Load Preset behavior, then verify every active level after loading. |
Use Destiny Inscription to repair the failure you actually observed. Side Bag access solves inactive-storage pressure; Home Repairs leads into the town progression used for Talents; other nodes can change starting resources, recovery, rerolls, or inventory capacity as shown by the live tree. Do not copy an Early Access purchase order or a cost from an old screenshot. Follow the current prerequisite lines and quest markers.
When Home Repairs is available, follow its current quest text rather than memorized coordinates. Version 1.0 fixed a case where the quest could complete without its conditions and added markers for certain quests. After the repair, inspect the available Talents and choose one that addresses the recorded failure. The important progression is not “always buy defense first” or “always rush damage.” It is that the permanent upgrade and the next expedition test the same problem.
Keep the weapon branch and the strongest working interaction for the next attempt. Change the earliest failed decision—node purpose, unsupported Artifact, Tablet orientation, movement reserve, or town upgrade—and see whether the run reaches the next checkpoint with more options intact. That is faster evidence than replacing the entire build with a launch-week tier list.
Evidence used and cutoff
Game identity, Steam AppID 2436940, TEAM HORAY credits, current feature scope, and original screenshots were verified through the official Sephiria Steam page. The official 1.0 update and 1.0.21 hotfix establish the July 31 launch boundary, inventory-system changes, current content scale, Preset fix, and balance-sensitive warnings used here. Readable 2025 beginner references were used only to cross-check durable interaction concepts; their exact Artifact rankings, Talent priorities, costume values, unlock counts, and weapon routes were not treated as current 1.0 facts. Removed, incompatible, incomplete, or internally conflicting pages were excluded from factual support.