← Rogue LabUser guide

Rogue Lab mechanics notebook

Status: prepared protocols; no in-game observations recorded.

Use these protocols to investigate specific unresolved mechanics. Description estimates, pinned export values, and calculator outputs are hypotheses, not trial results; notebook entries do not change calculations. Artifact descriptions are v57; tower/hero and matching cooldown exports are v56.0, revision f818c39e6d7cc4e0f8fb02a2e3d1a929bcfe6a24. Game version and platform must be recorded independently for every session.

Research before scheduling gameplay trials

Review official versioned patch notes and pinned gameplay fields first, then seek reproducible community measurements or recordings with identifiable versions and loadouts. Record source-backed conclusions separately from observations made by this project. Description repetition and undocumented videos do not settle stacking formulas. Use gameplay trials for remaining questions that can be isolated with reproducible controls.

Source review on 2026-11-12 (no project gameplay trials):

QuestionEvidence and current conclusion
Dreadbloon category gateOfficial v49 notes establish tier-4 towers / level-15 heroes. The calculator uses those thresholds; other conversion interactions remain unverified.
Divine Intervention overlapThe same v49 notes fix other crit artifacts overriding Divine Intervention. They do not establish joint probabilities, damage composition or acquisition-order independence. Overlapping artifact crit estimates remain excluded.
Malicious Mauling ability penaltyThe same v49 notes establish an ability-cooldown penalty, but do not settle its exact factor/scope against the older export. The ability clause remains excluded; see docs/ability-cooldowns.md in the repository.
Druid vines and critsOfficial v53 notes exclude artifact crits on target-destroying x3x vines. Pinned source identifies the destruction component. It and unmodeled spawned ground-vine output are excluded from ordinary hit/DPS estimates; see docs/tower-stats.md in the repository. Old demonstrations can show behavior that was later fixed.
Universal attack/ability cooldown floorsThe initial public-source search did not establish a reliable universal rule. Floors remain unknown; no absence-of-a-floor or guaranteed uptime is inferred.

Shared controls

  1. Record game version, platform, mode, map, encounter/round, game speed, recording

frame rate, tower/path or named hero/level, targeting, placement and target.

  1. Record the complete artifact loadout with exact rarities and acquisition order,

curses, shop bonuses, Monkey Knowledge and all external buffs. Mark unknowns.

  1. Use one isolated repeatable attack component, no innate crits or extra damage

sources where possible, and a target it can visibly hit. Confirm contacts in the recording; calculator hit eligibility is still unknown. Keep geometry, distance, target properties and non-tested effects fixed.

  1. Record a no-tested-artifact baseline and each tested artifact alone before a

combination. Repeat the baseline after a series to detect drift. If the game cannot reproduce the same conditions, label the comparison confounded rather than silently treating different encounters as controlled trials.

  1. Save raw video/screenshots and per-hit or per-cast data with timestamps/frame

numbers. Report misses, interruptions, target deaths and discarded intervals with reasons. Never infer per-hit damage solely from a rounded DPS display.

  1. Use independent repeated trials. Report sample sizes and spread; a short run

without an event cannot establish that event is impossible. Stop or redesign if an isolated observable component or legal loadout cannot be obtained.

CRIT-01 — artifact critical stacking

Question: when two applicable artifact crit mechanics are present, do their chances/damage interact, replace one another, or depend on acquisition order?

Candidate setup: a 400 Sniper with One Shot, Critical Success and Essence of Dreadbloon as the source-described high-tier category conversion. Check the actual descriptions and component for the chosen version and rarities before testing. Conversion behavior is itself unverified; hold the same conversion in every arm, verify each crit alone, and label the test confounded if those controls cannot be established. These are test candidates, not claims that a particular combination works in the game.

  1. Choose and record one legal setup in which each tested crit is independently

observable. Copy the exact rarity descriptions and their evidence links from Reference; do not substitute assumed multipliers.

  1. Record four arms with the same tower and other effects: neither A nor B, A

only, B only, A+B. Repeat the combination with B acquired before A. If baseline restoration is unavailable, use separately documented matching runs.

  1. For each hit, record index, frame, observed damage, normal/crit classification

(or unknown), and any relevant visual indication. Retain large hits and any possible simultaneous proc as raw observations, without forcing a model.

  1. Preselect a practical count per arm and repeat across runs. Compare observed

hit-size classes, frequencies and sequences with explicit hypotheses: independent events, exclusive events, replacement, or order dependence. Separate changes in normal-hit damage and attack interval from crit behavior.

  1. Report uncertainty and any indistinguishable hypotheses. Do not enable combined

critical expectation from a mean damage measurement alone.

Current calculator behavior: multiple applicable artifact crit mechanics warn and exclude the critical estimate. Innate tower crits remain excluded.

RATE-01 — attack interval and possible cooldown floors

Question: do measured attack intervals keep following increasing compatible speed modifiers, or is there a repeatable lower limit?

  1. Choose a simple direct attack without charge-up, changing attack sequences,

extra projectiles or conditional speed. Record its component and unmodified interval. Verify it is actually firing continuously at the same target.

  1. Prepare an ordered series: baseline, one applicable speed artifact, then

additional known applicable speed modifiers or separately controlled rarity replacements. Record complete arms and description/source expectations before measuring; a stronger rarity replaces the earlier copy.

  1. Count emission frames over many full intervals: `(last frame - first frame) /

(frame rate * number of intervals)`. Also retain individual intervals, game speed, recording precision, sample count and spread. Exclude wind-up and target reacquisition explicitly; keep discarded data in the record.

  1. Repeat each arm and the baseline. Change recording resolution or choose a

slower baseline if expected differences approach a frame. A plateau at frame precision can be a measurement limit, missed emissions or animation limit.

  1. Only propose a floor if observations distinguish it from those alternatives.

Report the tested component/loadouts and uncertainty; do not generalize a component-specific result to all towers.

ABILITY-01 — repeatable ability cooldown and possible floors

Question: how do applicable cooldown modifiers combine for one repeatable timer?

  1. Choose one ability with a repeatable timer and no extra charges, resets,

transformation or duration ambiguity. Record its exact identity and upgrade.

  1. Compare baseline, each cooldown modifier alone and combinations, keeping the

ability and other effects fixed. Do not mix placement cooldown rounds with ability cooldown seconds or attack intervals.

  1. Record cast, timer start and next-ready frames separately, with ability duration

and UI timer values. Count next-ready intervals instead of human recast delays. If readiness is ambiguous, mark it unknown and retain video.

  1. Repeat and report uncertainty as in RATE-01. A displayed integer timer cannot

establish a subsecond floor. Keep charge/reset behavior outside this test.

EFFECT-01 — ambiguous gates, conversions and aura effects

Question: write one specific unresolved clause from a bundled description. Examples include category-conversion order and an aura's provider/recipient gate. Detection, immunity and curse precedence require separate evidence; this protocol does not establish complete hit eligibility or survival feasibility.

  1. Copy artifact ID, rarity, exact clause and pinned evidence link. State at least

two interpretations and the observable outcome that would distinguish them.

  1. Define a control and one-variable changes: e.g. inside/outside the same aura,

below/at a described tier gate, or opposite acquisition orders. Keep unrelated categories, upgrades, targets and buffs fixed. Tier changes may alter base stats; compare each tier with its own no-artifact baseline.

  1. Confirm a provider exists before testing its aura. Record distances/placement

and acquisition order. If curse, conversion or another gate cannot be held constant, record that limitation instead of attributing its result to one effect.

  1. Measure the relevant observable directly and repeat. Record a failed trigger

as an observation with controls, not proof that every use of the effect fails.

Blank session / trial record

Duplicate this block for each session and attach raw data. Blank means untested.

Session ID / date / tester:
Protocol ID / question:
Status: prepared | observed | inconclusive | reviewed (choose one)
Game version / platform / mode / map / encounter / game speed:
Recording frame rate / timing uncertainty:
Tower / path or hero level / component / targeting / position:
Target properties / contact confirmation:
All artifacts / rarities / acquisition order:
Curses / shop / Monkey Knowledge / buffs / unknowns:
Source description and pinned links / calculator revision:
Calculator build report or build JSON (hypothesis only):
Hypotheses / predicted distinguishing observations:
Baseline and trial arms / intended counts / repeat count:
Raw file paths or links / timestamps:
Per-hit rows: arm, run, hit index, frame, damage, classification, notes
Per-interval rows: arm, run, start frame, end frame, interval count, notes
Exclusions / discarded rows and reasons / deviations / confounders:
Observed counts / values / spread / measurement uncertainty:
Interpretation / alternatives still consistent with observations:
Reviewer / review date / narrow supported scope:
Follow-up test / rule PR link (if separately approved and verified):

Review before changing a rule

Keep the raw record, version, controls and uncertainty available to a reviewer. Separate observed values from interpretations; conflicting or confounded trials remain inconclusive. Scope conclusions to the tested versions, towers and effects. Follow the repository's docs/artifact-rules.md and docs/tower-stats.md review procedures before implementation. Add regression coverage for the supported behavior and preserve all unrelated exclusions and unknowns. Notebook entries never automatically promote data or alter arithmetic.

← Rogue Lab