Version 0.12 · not yet playtested
This is the System Reference Document for Wizards & Lizards: the whole rules engine in one place, plus the guidance for building on it. The engine is small on purpose. It runs a fantasy delve and a deep-space salvage job with the same handful of dice, and it asks the table to describe what happens rather than look it up.
Everything here is released into the public domain (CC0-1.0). Take it, change it, publish your own version under your own name. No permission, credit, or royalty is asked or owed.
Wizards & Lizards is a Free Kriegsspiel engine at heart. Players say what their characters do in plain words. The referee decides whether it just happens, and when it does not, names the odds and calls for one roll. There are no target numbers to look up, no character build to optimize, and no stat blocks to consult. The fiction is the rules, and the dice are there only to find out what a genuinely uncertain, genuinely dangerous action costs.
That is enough to run a whole session out of the box, from a single starter page: a setting plus the barest hint of mechanics. From there the game grows one short supplement at a time, adding whatever a table wants (playbooks, travel, encumbrance, factions) until it is as heavy as that table cares to carry. Nothing is required except the engine and a setting to point it at.
The system is built in layers, and almost all content sits on top of the engine, never inside it.
The rule that keeps the system coherent: when you are unsure where something belongs, push it out of the engine and into a layer above it. A new archetype, a new kind of magic, a new subsystem: these are supplements or starter content, not engine edits. The engine only changes when something becomes truly universal to every table in every genre.
These chapters are terse reference. They state the engine plainly and lay the design dials bare, for whoever wants to author: a starter page, a supplement, a creature, a magic tradition, a whole setting. The final chapter is a starter-page template, the neutral skeleton an author fills in to aim the engine at a new world.
A note on dice: every roll is a pool of one to five d6, and you read the two best. How many you throw is the position the referee names; a 5 or 6 is gold, and how much gold you show decides success, a partial, or failure. Nothing is added to the dice. The next chapter is the whole of it.
When an outcome is genuinely uncertain and failure would cost something, the referee names your position, you roll that many d6, and you read the two best dice. Otherwise the action simply happens; do not roll.
A position is a pool of one to five dice. That pool is the whole of your standing: your skill, your footing, and the situation, folded into one number the referee says out loud.
| Position | Pool |
|---|---|
| Desperate | 1d6 |
| Poor | 2d6 |
| Fair | 3d6 |
| Good | 4d6 |
| Commanding | 5d6 |
Roll the pool and take the two highest dice. (At Desperate you have only one die; pair it with a 1 for the missing second.) A 5 or 6 is gold. Read the two you kept:
| The two best | Result |
|---|---|
| Two gold (two 5-6) | Success. You do it cleanly. |
| A 1-2 and no gold | Failure. It goes wrong; the referee makes a hard move. |
| Anything else | Partial. You do it, at a cost the referee names. |
Nothing is ever added to a die, and the pool never exceeds five or falls below one.
| Position | Pool | Success | Partial | Failure |
|---|---|---|---|---|
| Desperate | 1d6 | 0.0% | 33.3% | 66.7% |
| Poor | 2d6 | 11.1% | 55.6% | 33.3% |
| Fair | 3d6 | 25.9% | 59.3% | 14.8% |
| Good | 4d6 | 40.7% | 53.1% | 6.2% |
| Commanding | 5d6 | 53.9% | 43.6% | 2.5% |
Three things follow from this and are worth stating plainly:
This is the referee’s core craft, and it replaces every difficulty number in other games. Start at Poor (2 dice, the neutral unaided default), shift by the character’s attribute for the roll (a +1 lifts it to Fair, a −1 drops it to Desperate), then add a die for each thing genuinely in their favour and drop a die for each thing against them. The pool is capped at five and floored at one, so a well-set-up character has room to climb from Poor all the way to Commanding.
Add a die:
| Factor |
|---|
| The obvious right tool, or a specialist tool used as designed |
| Surprise, height, cover, or reach held |
| Real preparation: the groundwork was actually laid |
| An ally genuinely assisting |
Drop a die:
| Factor |
|---|
| Makeshift, improvised, or the wrong tool |
| Wounded (see Harm and Danger) |
| Working in the dark, underwater, one-handed, or under press |
| A foe or obstacle plainly beyond an ordinary person |
| Outnumbered (more foes on you than you can meet; badly outnumbered drops two) |
A background (see Characters) grants permission: if a task follows plainly from what a character used to be, they attempt it at their normal position without argument. Working outside a background is not forbidden; it is simply done at a worse position, the way anyone would.
You do not need to know how hard a lock is. You name the position, and that is the entire ruling. A hard task with a great approach is Good; an easy task done badly in the dark is Poor. The difficulty is already inside the name. Arguing the position up, out loud, before the dice leave the hand, is the most effective thing a player can do: every die you talk your way into is one you can see in your hand.
Four attributes describe what a character is made of. At creation, spread −1, 0, +1, +1 across them.
| Attribute | What it covers |
|---|---|
| Might | force, endurance, hauling; also how much you can carry |
| Agility | finesse, stealth, climbing, ranged attacks |
| Wits | noticing, knowing, tinkering, field medicine |
| Will | nerve, command, bearing, magic |
An attribute is a die. The neutral, unaided roll is Poor (2 dice); an attribute shifts that starting pool before anything about the situation is counted. A +1 lifts it to Fair, a −1 drops it to Desperate. So with the −1/0/+1/+1 spread a character starts a roll at Fair in their two strong attributes, Poor in their middling one, and Desperate in their weak one, then tools, position, and help carry them up from there. The −1 is where a character is a person rather than a build.
A setting may rename an attribute for flavor (Agility to Finesse, Will to Nerve), but keep the four-pillar split.
Stress is nerve and effort spent under pressure. Every character has five stress boxes. Mark stress to cast a spell, to push the body past what it should do, to buy a reroll (below), or when the referee says an effort costs it. At five marked, further strain wounds the body instead (see Harm and Danger). Stress clears only on a full night’s rest somewhere safe.
Rank is the only advancement the base engine carries, and it grants no bigger pool. (A heavier rung may add a defined arc that does; see the Playbooks supplement.) Everyone starts a Novice.
One reroll per roll; one die out of the pool; none of it at five stress. Because you keep the two best and success wants two golds, a reroll that turns a middling die gold can lift a partial to a success, or a failure to a partial. Rank costs no action and is paid after the dice have spoken, the one moment certainty is worth buying. A starter or supplement decides when rank is awarded; the engine only fixes what a rank does. In the base game, characters grow in fiction, permission, and reach, not in the size of the pool; a rung like Playbooks is where a defined, pool-growing arc is added for a table that wants one.
Characters have no hit points to whittle away. They have a condition, and it only ever gets worse:
HEALTHY → WOUNDED → INCAPACITATED → DYING → DEAD
Harm reaches a character only through their own roll (a “wrong” result, or a “cost” result whose cost the referee spends as a wound) and through hazards and clocks the fiction spells out. The referee never rolls to hurt anyone.
Every wound is one step down the track, whatever dealt it; there is no soaking a blow by breaking your kit. The one thing that turns a wound aside is worn armor: a suit ignores the first step of harm a character would take in a fight (a scene). After that first step, and for anyone wearing none, harm lands on the track.
Armor does not wear out, does not stack, and is not a resource you spend down: you get the one negation from the one suit you wear, it comes back each fight, and there is nothing to hoard for it or run dry of on a long road. What keeps a character alive is a good position, so the blow never lands, and armor for the one that does, not a pack full of breakable shields.
To stabilize a dying ally, someone who can reach them marks 2 stress or destroys a healing supply (bandages, a draught, a relic). It always works and never makes things worse. The clock is erased and the ally rests at Incapacitated. Reaching them is the part you roll for, if the way there is itself uncertain and dangerous; the clock ticks whether or not you get there this exchange.
A rescuer pays what they can: mark whatever stress is left, and anything still owing comes out of the body instead, one step of harm. A character at five stress can still save a friend, and will be Wounded for it.
The only routine healing is rest. A full night somewhere safe, with food, clears all stress and walks a character one step back up the track.
Come back from Incapacitated or Dying and a character does not come back whole. They take a lasting hurt, named by the player: the knee that locks in the cold, the ear gone deaf, the cough that starts whenever they run. One hurt per time they go down, however far down they went.
A hurt costs no dice; the track already charged for the wound. What it costs is room: it takes a slot in the pack, where it sits and where nothing stacks on top of it. The character carries it instead of something else, until someone who knows the work puts it right (coin and time in a town). Some never come right.
There is a floor. When hurts fill half a character’s slots, rounding up, that character is finished: not dead, but retired. They cannot carry what the work needs, and everyone can see it. This is the body’s own way out, and it is the one nobody chooses.
Combat is not a separate subsystem. It is the engine, run at knifepoint. There is no initiative and no turn order. A round is one exchange: everyone who acts rolls once, and the referee narrates the results in whatever order makes sense.
The one roll covers everything. Never roll again for damage.
A foe’s threat arrives through the player’s roll and through nothing else. Both a Partial and a Failure cost you the same kinds of thing; what separates them is whether you hurt the foe on the way. A Partial fells a foe as surely as a Success does, only bloodier, which is why fights here are quick and dear.
Fights are won before they start, with a collapsed floor, a jammed door, a fire, a bribe, or a corridor narrow enough that one foe reaches you at a time. Walking up to a large thing and swinging until one of you falls over is how parties end.
Creatures have no stat blocks and no hit points. Everything they do arrives through a player’s die. An author needs two things: a tier, and one true sentence.
| Tier | Blows to finish | What it costs them | Facing it |
|---|---|---|---|
| Lesser | 1 | One step of harm | Fair |
| Common | 3 | One step, and 1 stress | Fair |
| Greater | 6 | One step of harm | Poor |
| Dire | 8 | One step, and a clock fills | Poor |
A blow is a Success or a Partial against the creature; a Partial fells it just as well, but you pay for it (a Success is the only free one). A blow lands on most rolls, so a party overwhelms a lone foe fast, which is why the counts run high: a Lesser drops in one blow, a Dire soaks eight. The Facing it column is the pool the tier leaves a trained combatant (a +1 in the attack attribute), before anything else: a Greater faces such a fighter at Poor because it is faster, better, or larger, and Lesser/Common leave them at Fair. A less apt attacker rolls a die lower (their weaker attribute), an abler or better-positioned one a die higher, and a party that wants a fair fight must claw dice back with fire, numbers, height, or a doorway. A person’s pool can be driven to the floor of one die (a wounded fighter facing a Dire foe in the dark), where a Success is off the table and it is two-to-one they simply fail; hand them the die and let them decide to run. Never announce the odds; narrate the state (it is limping, it is not slowing down at all).
Outnumbered. A lone foe is weak, because a party of four lands four blows to its none. Numbers are how weak things kill: when a character faces more foes at once than the party can meet, they drop a die (they are Poor where they would be Fair); when they are badly outnumbered, two or more to one, they drop two. This is the teeth behind there are too many of them: a swarm is deadly not because any one of them is tough but because it forces the whole party to fight at a worse pool until they thin it, block the corridor, or find something that hits many at once. The penalty eases as the party fells foes and worsens as the party goes down.
The one true sentence does the work other games hand to keyword lists. Rule consistently from it:
This “harm sensitivity in one plain sentence” is the engine’s whole answer to damage types and immunities. It is always telegraphed, so it reads as a puzzle with a stated solution, never a hidden wall.
A character is four attributes, a line about who they were, what they carry, and one weakness they lean on when the pressure is too much. There is no build to optimize and no class to pick. The genre and the specifics come from the starter page; this chapter is the shared skeleton and the guidance for authoring the pieces a starter fills in.
Spread −1, 0, +1, +1 across the four attributes (Might, Agility, Wits, Will; see The Engine). A stat is a die, not a number added: a +1 adds a die to the relevant roll and a −1 removes one, before the situation is counted. Nobody is good at everything, and the −1 is where the character is a person instead of a build.
But every starting character is good at something. The neutral roll is Poor, and the two +1 attributes each lift a roll to Fair, so a hero is better than Poor in at least two of the four modes of play; the right gear or a signature tool lifts a third besides. No starting hero is stuck at Poor across the board, whether the edge comes from stats or from kit. (The one deliberate exception is the level-0 funnel, where the cast are ordinary folk, Poor or worse at everything, and survive by numbers rather than skill; see the Funnel supplement.)
Write one line: what the character was before this. Poacher. Ship’s surgeon. Deserter. Disgraced financier. Corporate sapper. A starter may offer a table to roll on.
A background is permission. If a task follows plainly from what the character used to be, they may attempt it at their normal odds, and the referee will not ask which box was ticked. It opens doors without closing any: working outside a background is never forbidden, just done the hard way, at worse odds, the way anyone would. This replaces a list of special abilities. Name who the character is, then play that person.
A character starts with a small loadout. Gear does what it plainly does, and carries no keywords or statistics. A crowbar pries and it also cracks a skull. A specialist tool used as designed is worth a die or two on the roll it was made for (see Naming the Odds); that is the whole of gear’s mechanical weight.
Worn armor is the one piece of kit the harm rules single out: a suit turns aside the first step of harm a character would take each fight (see Harm and Danger). It is not spent, stacked, or stockpiled, so armor is a standing bit of toughness, never a pile of breakable hit points.
A starter offers a handful of ready loadouts (three or four is plenty) plus room to assemble a custom one. Each loadout should imply a way to play, not a set of bonuses.
A character carries 10 slots plus their Might. One slot holds one weapon, one armor, one book, one light source, or one bundle of small things. Big things take more slots; a lasting hurt takes a slot and nothing stacks on it.
You can cram more in: a slot two deep drops a die from everything physical, three deep drops two dice. The penalty is the deepest stack on the sheet, not the total, so a party spreads the load or someone plays pack-mule at worse odds. Light and food are spent by the push (one meaningful advance), which is the clock the whole game runs on outside a fight. A table that wants carrying to matter more can swap this deepest-stack rule for the heavier item-bulk and load-band model in the Encumbrance supplement.
Name one vice: the thing a character takes to get the shaking to stop. A vice clears some stress in the field (the only thing that does, short of a safe night) and always costs something in return, a die off the next roll or worse. There is no version where the vice is the healthy option. A starter names the vices its world offers and what each costs.
Rank is the engine’s only advancement, and it grants no bigger numbers (see The Engine: Novice → Expert → Master, buying rerolls with stress). What the engine leaves to the starter or a supplement is when rank is awarded. Pick a trigger that fits the fiction:
Let players claim a rank when they believe they have earned it; if it is unclear, ask the table and take the answer. Because rank never inflates the dice, awarding it a little early or late costs the game nothing.
When writing the character section of a starter, supply:
That is a complete, playable cast with no archetype list and no point-buy. Heavier character options (playbooks, advancement trees, skill lists) are the business of supplements on the pamphlet ladder, never the engine.
Magic runs on the same engine as everything else. A setting decides what magic is (grimoire sorcery, divine petition, psychic talent, corporate wetware), but the mechanical spine below is shared, and an author changes the color, not the machine.
Magic lives in a source the caster must hold or hold to: a spellbook, a reliquary, an implant, a focus. By default a source takes a slot in the pack, and anyone holding it may attempt what it contains. A one-use source (a scroll, a charge) works once and is spent.
To cast: say what the magic does, mark 1 stress, and roll the odds with Will as the stat (a setting may rename Will). Read the usual ladder, with one twist that gives magic its texture:
Because casting itself is a roll at whatever odds the caster has clawed together, even a well-positioned caster sees a good share of castings go wrong. That danger, not a spell-slot budget, is the brake. Marking stress to cast, and the five-box ceiling beyond which strain wounds the body, is the resource. There is no mana and no separate slot economy in the engine.
Every spell is written as three lines, one per outcome. This is the author’s whole job. The Success line is the promise; the Partial line is the same effect plus a bite; the Failure line is the effect turning inward or a fresh danger. Rule something better suited to the room whenever it suggests itself; the lines are a default, not a wall.
A worked model, in a grim-fantasy voice (swap the color per setting):
| Spell | Success | Partial | Failure |
|---|---|---|---|
| Mend | A wound closes; one step back up the track. | It closes, and the caster gains a stress. | It closes wrong; the target takes a step down. |
| Bolt | A beam of harm: a step, at range and through worn armor. | The step lands, and the caster’s arm goes numb until rest. | It goes off in their hand, or into whoever is nearest. |
| Grasp | Roots or force pin a foe; everything against it two dice better. | Held a moment, still acting. | The grasp takes the caster or a friend. |
| Veil | Unseen and unheard for a minute. | Concealed, and everyone under it gains a stress. | The veil bursts; every eye turns. |
| Ward | A foe’s gear fails: armor, blade, or hinge, their choice. | Their gear seizes for a moment. | The caster’s own gear degrades. |
To give a setting its own magic:
Heavier magic subsystems (a component economy, prepared rituals, anchored lasting effects, a corruption track) belong on the pamphlet ladder as optional supplements, so a table adds only the weight it wants. The engine asks for nothing beyond stress, a source in a slot, and three lines per spell.
The referee’s job is to describe a living, dangerous situation honestly, decide when an action is uncertain enough to need dice, name the odds, and spend the results. The players roll; you never roll against them. Everything below is the craft of that.
Dice come out only when an outcome is genuinely uncertain and failure would cost something. Do not roll for what a competent person simply does. Do not roll when the fiction has already answered the question: oil poured and lit means the thing burns, no roll. Say no to the impossible before dice are ever involved. Most of play is description and rulings; the dice are the exception, not the metronome.
Name the odds out loud, with the reason, every time: “that’s Poor, two dice, you know this work but you’re doing it in the dark,” or “that’s Good, four dice, your best, and you set it up.” The reason invites the players to argue the pool up, which is the game working.
When a player rolls a Failure, something happens. Not “you miss.” Pick the move the fiction is already pointing at:
A Partial uses the same list, but they keep what they went for (and, in a fight, they hurt the foe). That is the only difference between a Partial and a Failure. Draw harm from the list at about the same rate on both: a Partial should cost blood roughly as often as a Failure does, so that climbing the ladder, which turns failures into partials, does not quietly turn the smart player into the bloodied one. Position buys the thing you went for, and buys you out of failure. It never buys a whole skin.
Draw segments for anything closing in and name it. The name sets the size and what fills it, and the fill condition is what makes the three clocks genuinely different lengths of time:
| Clock | Segments | What fills it |
|---|---|---|
| Imminent | 4 | Every Failure rolled in the scene, and every Partial’s cost you spend as time. Resolves this scene. |
| Looming | 6 | One per scene, plus one each time you spend bring the world closer on it. Arrives this session. |
| Distant | 8 | One per delve, journey, or week. A campaign threat filling over weeks. |
Scale a clock by what fills it, not by how many boxes you drew. Outright failure is uncommon once a party has any footing, so an Imminent clock leans as much on the time a Partial costs as on failures: a scene resolves because the party kept paying to get through it, not because they kept whiffing. The Looming clock’s dependence on bring the world closer puts the session’s threat in your hands, not the dice’s: spend the move and the players hear the world take a step. Searching a place, being loud or slow, and resting somewhere unsafe fill clocks too, which is the cost that keeps a party from turning out every wall.
When a clock fills, the thing happens. No roll saves them. Clocks make time itself dangerous without a die ever being rolled. (The dying clock in Harm and Danger is not one of these: it runs on exchanges alone, so the table can count it.)
Clocks also run the other way, as progress you fill for yourself: a project, a piece of research, a stronghold work. A progress clock fills on your own rolls (Success two segments, Partial one, Failure none) rather than on the world’s moves, and when it fills, the thing is done. It is the same tool pointed at what you are building instead of what is closing in; the Downtime supplement runs on it.
On a bad roll during exploration you may spend a resource instead of hurting anyone: a torch, a ration, a charge. Attrition is a slower, meaner threat, and it works.
A campaign is a string of expeditions with limited recovery between them. Its difficulty lever is the rest gate: whether the party gets to hole up and clear its tracks, or is pressed straight on. This, not any single stat block, is what makes a hard session hard.
For a campaign threat, keep one Doom: a clock apart from the three above that ignores their fill conditions and advances only when one of them runs out. It is the slow certainty the whole campaign leans against.
Award rank on the trigger the starter sets (see Characters). Heavier campaign structure (downtime turns, travel procedures, strongholds, factions) lives in supplements on the pamphlet ladder, so a table adds only the campaign machinery it wants. The engine needs none of it to run a session tonight.
The engine has almost no numbers, so how hard the game is comes from how you use the few dials below, not from stat blocks. They are listed most powerful first. A few are baked into a setting by whoever writes the starter page; the rest a referee turns in prep and in play. This section is for both.
How often a Partial costs blood (referee, live; the single biggest dial). A Partial always costs something, but you choose whether the cost is a wound or something bloodless: footing, light, time, a spent resource, a plan gone wrong. Make nearly every Partial a wound and combat is roughly twice as lethal; spend most Partials as position or time and the same fight is survivable. Vary it by scene: bear down in the set-piece, ease off in the skirmish. Nothing else you do moves the danger as far.
The rest gate (referee, per arc). Whether the party can hole up to clear stress and heal a step between fights is the campaign-level lever. The same string of encounters is a manageable delve or a death march depending only on whether a safe rest is reachable. Deny it (continuous assault, a filling clock) to make an arc hard; grant a defensible haven partway to ease one.
How generously you name the odds (referee, live). You set the pool. Reward a good plan, the right tool, surprise, or a held doorway with a die, and a single extra die is worth more than any gear. How freely you hand out advantage is a steady difficulty setting, and players arguing their position up is the game working, not a loss of control.
How many foes, of what tier (both). A lone foe is weak; numbers are the threat (see the outnumbered rule). Weight an encounter by count first, tier second: a knot of Common foes, or enough Lesser ones to outnumber the party, is deadlier than one Greater. Save a lone Dire for a boss the party is meant to meet with a plan, not a fair fight.
How available armor is (setting). Worn armor, the one cushion, sets a floor under survivability: whether characters routinely have a suit, and whether relic armor that turns aside two wounds exists, is a large swing. A starter that arms its cast is gentler; one where armor is rare, bulky, or forbidden is harsher.
When rank is awarded (setting). Rank buys the clutch reroll. Hand it out sooner and the party is tougher down the stretch; tie it to a hard-won goal and the early game stays dangerous.
For starter authors: your page’s built-in difficulty is mostly (5) and (6) plus the foe roster you ship. State the setting’s stance on armor and the rank trigger, and give a roster with a spread of tiers, and a referee has what they need to set the weight of any scene with (1)-(4) at the table. Say so plainly on the page if your setting is meant to run lethal or forgiving, so the referee knows how hard to lean on dial (1).
A starter page aims the engine at one world and is playable out of the box. It is the smallest complete publication in the system: a premise, the character options for that world, the barest reminder of the dice, and a table or two to improvise from. Everything mechanical it needs is in this SRD; the starter only supplies the color and the specifics.
This chapter is the neutral skeleton. Copy it, fill each section for your setting, and cut anything the setting does not need. The whole thing should fit a short pamphlet. Keep the rules reminders tight and let the setting breathe.
A few sentences that say what world this is, who the characters are, why they stick together, and what they spend their time doing. Give the table an engine for scenes, not a lore dump. Name the tone in a line so the table pitches it right.
A short reminder, not a re-teaching (the SRD carries the full rules). Include:
Per Characters, filled in for this world:
A short bestiary in the engine’s shorthand: tier + one true sentence each. Four to eight is plenty for a starter. Seed at least one telegraphed puzzle (the “one true sentence” that names its own solution).
A starter bakes in a few of the difficulty dials from Running the Game (Tuning the Difficulty); state them so the referee knows the setting’s intent:
A hook engine so a referee can start cold. The classic is a d6 to find work (nothing / a job with a catch / a choice of two) plus a premise table: “[Name] wants [thing] at [place], but [twist].” Tailor every entry so the characters’ options come into play.
In the tradition of the micro-RPGs this system descends from, a back page of four tables of twenty lets a referee improvise a session: contacts, jobs, places, twists, or whatever this world runs on. These are gameable lore, ready to drop into play.
# <SETTING NAME>
<one-line pitch and tone>
<the premise: who you are, why you're together, what you do>
## The Dice, in Brief
<positions; ladder; name-the-odds; stress; condition track; sacrifice>
## Making a Character
Stats: spread -1, 0, +1, +1 across <four attributes>.
Background: <d66 table> — your permission and your hook.
Kit: <3-4 loadouts>.
Vice: <options + costs>.
Rank: <when Expert / Master are earned here>.
What you came for: <the goal line>.
## Threats
<4-8 entries: tier + one true sentence each>
## Getting Into Trouble
<d6 find-work> + <premise table>
## Back Page
<improv tables on d6: contacts / jobs / places / twists (a d66 list, or roll four d6 and read across the columns)>Keep the engine reminders short, put the world’s voice everywhere else, and the starter will do the one thing it exists to do: get a table playing tonight, from a single pamphlet.