Quick Guide
  • Open Combat Settings with C and change the Health Potion percentage. Wikily reports a per-character 0–100% slider with a default of 50%, triggering when HP falls below the selected share of maximum HP.
  • An earlier trigger creates more recovery time before the next hit, but can spend charges sooner. The potion has a five-second cooldown and heals partly over three seconds, so no percentage guarantees survival.
  • The launch-player suggestion of 75% is a comparison to test, not a universal optimum. Check the potion tier, remaining charges and refill system alongside timing.

Change the HP setting and confirm the right character

Open Combat Settings

Press C or use the Combat Settings panel. Find the Health Potion Auto-Use percentage rather than the Combat consumable assignment slot.

Record the starting value

Note the current character, maximum HP, potion tier and remaining charges. The captured reference stores the percentage separately for each character.

Choose one timing change

For late recovery, compare an earlier threshold with the previous one. If testing 75% against 50%, keep the map, equipment and stock unchanged.

Observe a complete fight

Watch whether a drink starts, whether the next damage arrives during the cooldown and whether the delayed healing has time to help.

Follow the refill cycle

Check charges after town or revive. Decide using repeated survival and charge consumption, not one lucky boss result.

The wording is “below” the chosen share. A number on the slider is not a promise of an instant action exactly at that HP value, nor a guarantee that a single hit cannot cross from safe HP to death. If no drink happens, first distinguish an empty charge pool, a cooldown and a damage spike from a setting that failed to save.

Know what one charge actually buys

Wikily describes three base charges plus Bonus Potion Charges. One drink spends a charge and starts a five-second cooldown. Each potion tier applies an instant fixed heal followed by a percentage of maximum HP over three seconds. The delayed part is not all available at the moment you trigger. That difference is central when an enemy hits hard again immediately.

Wikily — healing tiers and level gates; delayed percentage applies over three seconds
Potion tierHero levelInstant healDelayed share of max HPGold for upgrade to this tier
Minor125 HP30%Starting tier
Health575 HP30%100
Greater15150 HP35%500
Superior25300 HP35%3,000
Supreme35600 HP40%25,000
Divine451,000 HP40%80,000
Mythic551,500 HP45%150,000
Eternal652,000 HP45%300,000

The gold column is one upgrade’s listed cost, not the cumulative cost of skipping directly from Minor to the final tier. Priests upgrade one tier at a time and require ingredients as well as the relevant level. Reaching level 25 does not by itself upgrade the potion to Superior. Read the currently equipped tier before deciding that the recovery percentage is wrong.

Selected early upgrade ingredients — check the priest’s actual offered tier
UpgradeMaterials reportedUseful planning implication
Health, level 510 Mushroom + 5 StarleafReserve the first recipe materials rather than spending every plant on temporary buffs.
Greater, level 1520 Mushroom + 10 StarleafThe next permanent healing step uses more of the early plants.
Superior, level 2550 Starleaf + 30 Oak LeafGold alone cannot complete this step.

A worked timing example without inventing incoming damage

Suppose maximum HP is 1,000 and the character uses the level-five Health Potion: 75 HP immediately, then 300 HP over three seconds. At 500 HP, the nominal 375 total would bring HP to 875 if no further damage or other healing occurs. At 750 HP, the same nominal total would exceed the 1,000 maximum by 125 under that static assumption. Actual combat can consume the healing as it arrives, so this is an explanation of headroom, not measured waste or a recommended setting.

The lower trigger preserves more missing HP for the heal but leaves less room for another damaging hit. The higher trigger starts recovery earlier but may use charges during small recoverable losses. Five seconds must pass before another drink under the reported cooldown rule. If the next hit kills the hero inside that window, raising the percentage further may be less useful than upgrading the instant heal, improving defense or choosing a more suitable node.

Apollo quotes a launch player preferring 75% instead of 50%. That gives a concrete starting comparison when the problem is a late drink; it does not establish a boss-specific or class-wide optimum. Keep the same potion tier for the timing test so an upgrade does not hide the effect you are evaluating.

Fix the charge supply separately from timing

Charge and recovery methods in the captured references
MethodReported resultLimit to remember
Priest’s HealFull HP and mana, every charge refilledRequires an actual town interaction unless the relevant automation handles it.
Level upFull HP/mana and all charges restoredDo not depend on frequent level-ups for a long same-level loop.
Auto Refill, 500 goldRefill charges after return to town or reviveDistinct from Auto Recovery and crafted buff inventory.
Auto Recovery, 50 goldFull HP on return to townIts health-restoration description is not the same as the refill node.
Charge pickupsAdd a charge up to maximumOnline loot is planned by the server; source drop rules are not a guaranteed live supply.

If the charge count reaches zero, a higher HP trigger cannot create another drink. Watch a full repeat-and-return cycle and confirm that the chosen refill route actually runs. If the hero revives, the death reference ties charge refilling to Auto Refill; full health and mana on revive alone are not proof that every charge has returned.

Diagnose the failure before choosing another percentage

A symptom-first recovery check
ObservationLikely questionNext action
HP crosses the setting with no drinkWere charges available and the five-second cooldown over?Inspect both before moving the slider again.
Drink occurs, then death almost immediatelyWas instant healing too small for the next hit?Check potion tier, defense and encounter difficulty.
Survival improves but charges vanish quicklyDoes the earlier trigger spend more per cycle?Compare complete runs and refill sustainability.
A second character behaves differentlyDoes it have its own saved threshold?Read that character’s C-panel setting.
Town trip heals HP but leaves empty chargesWas Auto Refill bought or Priest Heal used?Fix refill rather than HP timing.
Deaths seem randomWhat enemy and damage type appear in the log?Record the death notice while testing, then inspect the matching defensive shortage.

The Log in Combat Settings records death notices with enemy, damage and damage type according to Wikily. It is text without timestamps and is not part of the save, so copy the relevant observation during the comparison. This helps distinguish a critical burst from a simple empty-charge loop without claiming that a log line proves the entire combat formula.

Q: Should I always use 75%?

No controlled universal result supports that. It is a launch-player suggestion to compare when 50% triggers too late.

Q: Will a 100% threshold stop all deaths?

A threshold cannot remove the five-second cooldown, generate missing charges or guarantee survival against a lethal hit.

Q: Is a Health Potion upgrade the same as crafting a combat flask?

No. The charged potion has permanent priest upgrade tiers; Combat and Utility flasks occupy temporary buff categories.

Q: What should I change first: threshold or potion tier?

If the drink is late, test timing. If it fires but the instant heal is inadequate, inspect the available tier and defensive problem. Change one variable so the result remains interpretable.

Sources and scope

Source-reviewed guide for Deskrawl: Idle ARPG, Steam app 4623570. Named mechanics and figures are attributed to retained community references or dated official patch announcements. The editorial pass reread the cited source bodies and recorded excerpts and hashes. Publisher independence is not established for community data mirrors. No installed-client test, measured best build, or live game-version confirmation has been performed.

Review status
verified
Game version
Community snapshots collected 2026-10-04/05; official 1.0.0h and 1.0.0i are used only for their announced changes.
Checked
Method
Bounded first-release dossier review, including source roles, preserved conflicts and explicit limits. Recomputed retained source hashes, exact excerpts and line ranges and signed reader content. No installed-client or live-server verification.
  1. Wikily — Health Potion charges and trigger community

    Supplemental source body read in this revision: HP threshold slider, per-character setting, charge refill methods and online loot limitation.

    Checked

  2. Wikily — Health Potion mechanics community

    Supplemental community source: three base charges, five-second cooldown and instant plus three-second healing. Not a consumable auto-use slot.

    Checked

  3. Wikily — Health Potion tiers community

    Community-reported eight upgrade tiers, level gates, instant/delayed healing and per-upgrade gold and ingredient costs.

    Checked

  4. AFK Meta — Life Skills dependency table community

    Data labeled 3 October 2026: Auto Recovery, first Revive Cooldown Reduction, Auto Refill and Combat Potion Auto-Use prerequisites; separate utility and food unlocks.

    Checked

  5. Wikily — death, revive and return community

    Community-reported revive countdown, full HP/mana, conditional refill, entry-item limitations and return-to-town behavior.

    Checked

  6. Wikily — Combat Settings log community

    Community-reported clear and death notices in the C panel; log lines have no timestamps and are not saved, so they are evidence to record during comparison.

    Checked

  7. ApolloTrail — launch-week beginner setup secondary

    Launch-era operation and player reports: combat settings, the 75% HP suggestion, first Life Skills prices, buff interruptions and the enhanced Common weapon anecdote. Stack-size and automation-timing statements conflict with newer community data and are not treated as universal rules.

    Checked