What is confirmed
The safest starting point for a Rift Wizard 3 spell guide is the official product identity: Steam App 4366330. Its developer description says that spells and upgrades can be purchased at any time, which makes the spellbook an open construction problem rather than a fixed class tree. The same listing advertises more than 200 spells and 400 spell upgrades, but those totals are developer claims and were not independently reconstructed row by row for this review.
A pre-release explanation from development-team member k. hoops shows Spell Points being spent on spells and says each spell can receive up to two upgrades. The README for the accessibility mod Words of Power II, pinned to its Aug 29 commit, independently describes the same two-upgrade limit while documenting the spell shop interface. Neither source proves that every individual spell value shown in an older recording still matches the Aug 31 main branch.
Early Access makes a date label essential because the official Aug 31 Patch Day announcement changed many spells after the launch-period videos were recorded. One named example in that announcement is a charge change for Boiling Blood, which is enough to show why copying old figures into a supposedly current table is unsafe. This page therefore treats Aug 31, 2026 main as its evidence baseline and avoids unstated damage, range, charge, or cost values.
Filter by function
A practical spell search begins with the combat job that is currently missing, not with a claim that one school is always strongest. Useful editorial buckets include removing a priority target, affecting an area, repositioning the wizard, creating allies, controlling space, and absorbing immediate threats. These are navigation aids distilled from guide practice, not official categories supplied by the developer.
For every candidate, ask what board problem it solves on the next turn and what it leaves unanswered. A powerful single-target answer can still be a poor purchase when the spellbook cannot manage crowds, while broad damage can fail against the wrong resistance. This functional check keeps the decision tied to an encounter rather than to a name copied from a launch-week recommendation.
Movement, defense, and summons deserve the same inspection as direct damage because a run is not scored only by a spell's isolated damage line. A repositioning tool may preserve an action sequence, a summon may redirect danger, and a control effect may buy time for other resources. Those outcomes still need to be read from the current in-game text; the examples explain roles without assigning unsupported numbers or universal priorities.
Tags and search
Tags are most useful as links between spells, upgrades, and equipment rather than as a rigid taxonomy. A community answer about ally boosts suggests starting with an enchantment tag, checking conjuration for many minion-related effects, and searching spell text for phrases such as minion damage. That is player workflow advice with an unknown patch number, so it should be tested in the current interface and never rewritten as “all buffs are enchantments.”
Search both the base spell and its upgrade text because a useful tag or interaction may appear only after an upgrade. Record the exact wording visible in the game, then distinguish the base tags from anything an upgrade adds. This prevents a filter result from silently claiming that an unupgraded spell already has a property it gains later.
A trustworthy searchable record would include name, SP cost, visible tags, damage type, charges, range, upgrade names, and a verified-patch field. Any field not checked against the current build should say “not verified” instead of inheriting a value from Rift Wizard 1, Rift Wizard 2, or an undated wiki. This structure is more useful than a long list whose version and game identity cannot be audited.
Upgrades and coverage
The supported rule is that a spell can take up to two upgrades, not that every spell presents exactly four choices. One third-party guide describes a choose-two-from-four pattern, but the stronger sources used here confirm only the resulting two-upgrade limit. Until the current interface or an authoritative data export establishes the number of offered choices for every spell, that broader claim remains unresolved.
Damage coverage matters because the development-team introduction demonstrates enemies with complete resistance to a damage type. That example supports carrying answers that do not all fail against the same defensive profile, although it does not establish a universal pair of damage types. Inspect the upcoming enemies, current charges, positioning needs, and available alternatives before spending SP.
Upgrade evaluation should also ask whether a choice strengthens the same role or opens a missing role. Improving a reliable answer may be correct when the run already has coverage, while a new interaction can matter more when the spellbook has a clear gap. No fixed formula was found, so the defensible result is a repeatable question set rather than a hidden score presented as fact.
Build a versioned record
When documenting a spell, first confirm that the source is explicitly about Rift Wizard 3 and preferably links to App 4366330. Capture the source date, the stated or inferred patch context, and whether the evidence is official, creator commentary, community experience, or interface-adaptation code. A launch-era demonstration can remain useful for a broad mechanism while being unsuitable for a current numeric field.
Next, separate observation from recommendation in the record. “The interface displays this tag” is an observation, while “buy this spell first” is advice that depends on the rest of the run. Keeping those fields apart makes future patch review possible without treating an old creator preference as a permanent rule.
Finally, recheck any row affected by an official patch and leave the prior value as dated history only if it helps explain the change. The demo also needs its own label because the Aug. 31 announcement says it ends after Realm 5 and offers about one third of the full spell and equipment selection. A demo catalogue is therefore not evidence of everything available in the full Early Access game.
What is not verified
No reliable source in this chapter provided a downloadable, current, complete spell table through the Aug. 31 main baseline. The review also found no public current build number, official tag glossary, or independently reproducible set of every spell's stats and upgrades. Those are evidence gaps, not proof that a field is zero or that an older page is complete.
Search results frequently mix the first two Rift Wizard games with Rift Wizard 3, so matching a familiar name is not sufficient. Pages using unsupported mana or class structures were excluded, as were claims built from removed or inaccessible material. The right response to contamination is to verify the game, patch, and visible text, not to merge entries across sequels.
Use this guide as a system for selecting and recording spells, not as a promise that every current spell has been enumerated. Steam's 200+ and 400 figures communicate developer-claimed scale, while the filter workflow keeps decisions useful under changing balance. A complete database should only be published when its rows, version, and collection method can all be audited.