Sunday, July 5, 2026

Odd Goblin 0.7 Update

An updated version of Odd Goblin is live! 0.7 brings a fairly major rewrite of the rules for additional clarity, many tweaks to classes, and 43 sample monsters. Some of the classes and monsters had abilities that were either stronger, weaker, or nonsensical in the face of previous rules tweaks, so all of those have been ironed out and rebalanced.

The most significant rules change is stealing the concept of Mortal Wounds from Mythic Bastionland--needing to roll a save to keep standing after taking damage beyond your hp in most Oddlikes was always one step too fussy for me, so removing the roll from that does a great job to keep the dynamic but reduce the complexity--and also bring Odd Goblin slightly closer in mechanical compatibility to other Oddlikes. It effectively gives everyone a +50% hp boost, which has felt needed--it makes it much easier for player characters to die, while giving monsters a bit more staying power (which they were proving to need--my level 1-4 players recently ran off an Ancient Dragon with only one loss, something I thought was completely impossible). 

Monsters are especially fun to write for Odd Goblin--I try to make sure almost all of them have some sort of special twist or unusual situation they can put the players into. They should all be fairly easy to convert to any Oddlike system. 


There are also a few character classes that I loved conceptually but just weren't clicking, mechanically. Each of those have been reworked to grant a few more concretely helpful abilities while still keeping their original charm.


The full changelog can be found below.

Changelog

  • Removed the GM pdf, since the polished parts of it have been moved to the main document. The spell/treasure lists are still available on the google doc as a tab: Here

  • Added credits page, including a link to the itch page and a discussion discord

  • Rulebook now contains both pages meant for players (2-40) and GM-facing material (41+).

  • Increased font size of rule pages.

  • Rewrote most rules pages. Most rules did not change in practice, but many were rewritten to be more clear.

    • Added an explicit flow for resolving damage each combat turn. The effect is largely the same as before, but hopefully smoother to resolve

  • Changed some terminology

    • HP is now guard. Max HP is now health.

    • Added a notation for ability checks and saves: 6STR means a strength save of difficulty 6.

    • An attack that rolls over the target’s armor is a ‘hit’

    • An attack that rolls over armor and leaves the target with zero guard is a ‘critical hit’. Some monster abilities can critically hit any time they roll high enough

    • Attacks that fail to do the most damage in a round  but still hit are called assists.

  • A few old rules didn’t work as well after adding the ‘each round you only get hurt by your biggest attack’ mechanic, so did a pass cleaning those up

    • Every assist adds +1 damage to the biggest attack

    • Dual wielding now grants two attacks instead of upscaling damage.

    • Got rid of the distinction between actions and attacks. You now just get one move and one action, which can be an attack. The intent of giving a move, action, and attack was to encourage characters to take non-damaging actions in combat, but diminishing returns on attack damage already do this

  • Spells that did [sum] damage have been dialed back to match the new damage rules

  • Stole the concept of ‘mortal wounds’ from Mythic Bastionland. You no longer fall unconscious at 0hp, now you fall unconscious when a single attack reduces your health by half or more

    • This implicitly gives most monsters a health boost, which playtesting was revealing to be needed

  • Players always go first if they take the initiative in the fight. If they are passive or surprised, they go second. If ambiguous, only those who pass a WIL save go first.

  • The Resilient background now gives you a health bonus at levels 1 and 2, since the faster healing was proving rarely relevant.

  • Dwarves always know which way is north.

  • Got rid of the concept of ‘reactions’.

    • A few character abilities that were reactions can now just be done once per turn

    • Guide protect now must be before damage is dealt, but grants 1 armor against the attack

  • Fighter regains Upscale Damage at template C. Swap can be used on 0 guard enemies and protects allies from opportunity attacks. Whirlwind grants bonus points on next round.

  • Rogue backstab ignores armor on a sneak attack, and sneak attacks more easily. Upscale Damage upscales one handed weapons an additional time.

  • Noble’s Natural Leader ability applies to NPCs under them, or all allies if the noble spends their action giving orders

  • Dream Thief now puts people to sleep at half health, regardless of their attack strength

  • Devourer claw attacks now attack once per free hand

  • Removed the damage penalty to the gunslinger when making multiple attacks, but capped shots per round at level.

  • Beastfriend’s companion now gains health as they do. Human speech moved to B template, C template gains Enhanced Companion, raising the companion’s damage.

  • Bard gains spell dice at B,C,D. Song of Healing allows them to transmit spell dice to other characters. Song of Rest moved to A, Song of Bravery moved to B. Celebrity changed to [level]-in-6 instead of [level]-in-10 odds, since level no longer refers to total character level.

  • Hero’s extra attacks and extra damage traded place in the templates. Armored split into Unstoppable, which ignores armor movement penalties, and Armored, which grants 1 armor.

  • Pointer ability into Scentsight and moved it to template C. Removed Growl. Upgraded Pack Hunter and Painful Bite. Split Painful Bite into Painful Bite and Teamwork.

  • Chaos Warrior exoskeleton raises health. Extra heads grant spell dice. Replaced their abilities that grant bonus dice (boring!) with new ways to trigger mutations.

  • Sentient Ooze folded camouflage into mimic, moved mimic to C, and gained Harden at D--granting armor and the ability to solidify into a ladder/etc.

  • Nasty Little Freak’s Filthy now grants immunity from things trying to eat you. Scavenger now works like Guide’s Forage. Coward now allows them to run instead of fall unconscious. Bully forces adjacent enemies to make saves with disadvantage.

  • Sage spellbook mastery changed to allow a single reroll of spell dice, since it stacked weird with spell mastery. Sorcerer spell mastery changed to a reroll, as well.

  • Spellblade cantrip now only affects weapons, but grants the armor penalty bonus while carried. Smite and Slice Air damage both reduced. Shield can no longer be cast as a reaction.

  • Druid’s Commune With Nature moved to [dice] hours instead of concentration, but ends if the greater power is invoked. Polymorph duration scales with dice instead of being concentration. Fog Cloud no longer fully blocks caster’s vision, scales up in range, and may travel with the caster, and has duration set to concentration + [dice] rounds.

  • Enchanter’s Bottle can no longer target containers. The spell gets confused.

  • Historian’s knock range scales with dice.

  • Internist Miniature Sun now does 1d4/1d8 damage, Fireball damage reduced.

  • Kleptomancer Ranged Touch only allows for empty hands to be projected. Silence range enhanced by dice.

  • Mentalist Perfect Recall also grants the ability to perfectly draw memories. Circle of Truth range expands with dice.

  • Necromancer Darkness may be cast on an object. Rules for being unable to return to your body after Release Ghost. Bind Soul can target a corpse’s original soul.

  • Puppeteer’s True Puppetry health caps raised slightly.

  • Seer’s See Future always exhausts spell dice used to cast it.

  • Summoner’s Summoning Mark, the most nerfed spell of the most nerfed class, suffers a (final?) nerf--the mark always vanishes on use, and applying to a creature now costs a spell die. Return Home now requires 10 minutes of chanting and 1 spell die to trigger the return trip. Gate may randomly drift in a random direction and has a duration of concentration.

  • Telekinetic Flight speed scales with dice. Catapult damage lowered. Shield replaced with Control Water, Poltergeist replaced with Magnetism. Removed drawbacks from Empower Limb but lessened damage.

  • Added a few FAQ examples

  • Added 43 monsters to the main document

Tuesday, April 28, 2026

Elements of a Key 4: Rules for Key-Writing

 Alright, last time I read a bunch of keys from various modules and looked at what types of info they included and how they presented it. It's time to try to put all of this to practical use. I'm going to make myself a style guide for my solo-author releases.

I've seen a fair number of arguments over whether good writing is necessary for a RPG product. I'm a firm believer in "rpg books are tools, the play at the table is the art", but even within that context it seems obvious to me that good writing is preferable to thoughtless writing. The more interesting question is "how is good writing for a RPG book different than good writing for a novel/poem/essay?"
I'm going to start by listing my rules for myself, then put them to practical use; I rewrote a one(ish) page dungeon I made last year using these rules--I'll post both copies so we can compare the two and see how well this all works in reality.

Rule 1: Break these rules if you have a reason to, otherwise don't.
I believe that every mechanistic ruleset is more fun if it contains the seeds of its own destruction; a totalizing ruleset that gives you no exit ramp is irresponsible to create. Skills and taste are far more important to good RPG writing than any list of rules could ever be, so if your taste tells you that you're making your writing worse by following a rule, or if following a rule feels more like busywork than design, you should break it.
If skills are better than rules, why write rules? Skills can't serialize onto a blog post, for one thing, but even more importantly a set of hard rules is a useful lens for looking at your own writing more thoughtfully; you learn more by choosing to break a rule than by ignoring one.

Rule 2: Everything defaults to unformatted paragraphs.
No bullet points, no bolding. I have in the past been a big fan of bullet points, and am not categorically against bolding, but Part 3 of this series made me realize that they do both come at a cost of legibility. I'm no longer sure what bullet points do that paragraphs do not (other than indent, potentially), and bolding only works when used extremely sparingly--bolding more than one phrase per paragraph loses all of the benefits of drawing the eye to key phrases and instead creates a visual cacophony.
That said, all the various formatting tricks people use serve a purpose. Rather than neglecting to use formatting, my goal is to study the formatting styles others have used and try to replicate their effects in the structure of my text. Plain text paragraphs still need to serve as reference documents, and still need to follow patterns that the reader can use to find information during play with minimal friction.

Rule 3: Initial paragraphs are for information immediately apparent upon entering a key.
I talk a lot about Immediate Sensory Impressions in previous posts. The first paragraph of a key should generally contain a summary of those impressions. To use the framework of Landmark, Hidden, Secret, it should contain the key's Landmarks.
A goal of this ruleset is to write content that makes the GM's life easier. The initial paragraph should be something that the GM can use as the foundation of their initial description of a key as the players enter it; ideally it should be something that they can read almost verbatim under normal circumstances. Of course, non-normal circumstances are commonplace in RPGs, so the descriptions should not be full scripted box-text.

Rule 4: Later paragraphs describe one topic each, and that topic should be recognizable from the paragraph's first few words.
I really liked Lair of the Lamb's approach in our last post, where each paragraph bolded its subject matter--this makes it extremely easy to skim between keys for specific information. This follows the same philosophy, but forgoes the need for bolding in favor of predictable sentence structuring. I may fully embrace the Lair of the Lamb approach in the future, allowing myself one bolded phrase per paragraph, but for now I'm sticking to Rule 2.
For reasons that not even I understand, I've been using inverted indentation for a while now--the first line of a paragraph sticks out fully to the left while all other lines are indented. This actually works really well with this rule--the first few words of each paragraph stick out, making them easier to scan. The goal is to make the contents of each paragraph apparent from that bit sticking out.

Rule 5: The more secret information is, the deeper it goes into the key/paragraph.
To again use the framework of Landmark, Hidden, Secret, you start by revealing all the landmarks in the first paragraph, then move to the hidden elements that can be straightforwardly discovered through the players asking questions, then go to elements that require player fiddling to discover. It is likely that hidden elements will be described in play before secret ones, and the key should be laid out in loose order of expected use.
This can be fairly flexible in practice. To use the LHS example of a rotating bookcase--the bookcase is landmark, the fact that it is a secret door is hidden, the mechanism of triggering it is secret--the bookcase should definitely be mentioned in the first paragraph, but after that things open up quite a bit. You could add a paragraph detailing the scuff marks on the floor by the bookcase, then a second paragraph describing how to operate it, but you could also just make those sentences one and two of a single paragraph. You could even completely omit the scuff marks--they're already implied by the fact that the door rotates in the first place. These rules are not an excuse for writing to become rote, the needs of the specific elements being described still come first.
If a key contains little enough information that it makes the most sense to write it as a single paragraph, that's fine--just follow this same basic flow.

Rule 6: Keys contain information that can be revealed in the key itself.
A key exists to assist in the act of a GM describing the events that occur within a space in play; it should focus on providing information that can come out while the players are inhabiting that space. If there is a secret about a location that can not be discovered or interacted with at that location itself, it should be described in the key for the location where it can be discovered. If there is no way for the players to learn the information, it should not be included in the module.
One exception to this is information that is discoverable, but not at any keyed location; if you find a treasure in a dungeon that any priest of a certain sect will recognize as a long lost non-magical relic, you can describe the priest's reaction even though they aren't in the dungeon key itself. However, if a church of that sect is a keyed location in the module, you should put the information there.

Rule 7: Information should be presented in the manner in which the players can discover it.
Related to Rule 6, it is better to describe the way in which players can access information than it is to simply state the information. Instead of "this is the workshop where the artist created the cursed painting" describe the mess of art supplies, the stacks of progressively darker and stranger failed early drafts, the strange reagents ground into pigments strewn across the room. This is "show, don't tell", but even more so--if you don't come up with the concrete details that do the showing the GM will need to improvise them; make their life easier. Of course, the key should be described clearly enough that the GM can grasp the information in case the players investigate the key in an unexpected manner.
A rote way to help with this is to check for usage of "to be" verbs and replace them with something more active. This can be as simple as replacing "there is a painting of the king" with "the painting reveals the form of the king", but it's a good check to make sure that you're describing information that can actually be revealed in play. It's also just more engaging from a writing perspective.

Rule 8: Write without assumption of perspective.
In slight contrast to the previous rule, information should be presented without assumptions made to specific player positioning or action. You never know if the players will encounter a goblin guard by accident, via sneak attack, while invisible, escorted by an ally of the goblin, disguised, etc. Instead of "they attack the players" write "they attack intruders". Avoid the second person; initial paragraphs should never say "You see a painting at the edge of your light as you open the door"--maybe you arrived in this room from an unexpected direction via passwall or teleportation accident.
Doors in general are tough due to this--they're the one element I often omit in my initial paragraph, just because I have no way of knowing which door the players are entering a room from. When I do describe a door, I describe it as being on the east or north wall, not the far or right one. Even this is unsatisfying, because the GM will probably need to strip out all of that directional information and replace it with ahead/left/right descriptions to maintain the mazelike nature of a location.

Rule 9: If information about or from another key is available within this key, provide a parenthetical key and page link for easy reference.
If you refer to another location in a key, and some element of this key either can tell you or take you to that key, provide a link to the other key in the format of (X.XX, pg. XX). Page flipping is always a pain, so anything you can do to reduce the need for it is good. The most common form of elements that can tell you about other elements are characters--if a character is likely to mention another key, and is able to answer questions about it, you should provide a link for convenience.
That said, if a key mentions another key but is unable to give details about it, no link is usually needed. A lost traveler trying to return to their home should have a link to their hometown, but an explorer looking for a lost city is fine by itself; the explorer can not give directions to the other key, and does not know the details of what it contains, so there is no need for the GM to refer to it. Many exceptions obviously apply--if the explorer has a map of the city, a link becomes needed.
Reminder links can be added as needed as well--in the relic example from Rule 6 I might link to the church as a reminder just because it's reasonable that a player might be playing a priest of that religion or be able to learn of it through some other unanticipated method. This is on a case by case basis.

Rule 10: Singular mechanical stats may be put in a parenthetical, but anything longer than a phrase should be described in natural language.
Parentheticals interfere with reading if used unsparingly, but some concepts are genuinely short and mechanical enough that it makes sense to step outside of natural language for them. The most common reason I've found is weapon stats--it's less clunky to say "a longsword (1d8)" than to spend a sentence describing that longswords do 1d8 damage. Also reasonable is "a flamethrower (15' cone, 3d6)". Anything more complex than that, though, should simply be described as a sentence.
Skill checks/saves are one of my greater sources of uncertainty with this ruleset. I plan on generally writing for a specific ruleset (usually Odd Goblin), but supporting easy conversion to other systems. Weapons and health and spells are often cross-compatible enough as-is, but systems range dramatically in how they handle skills and saves, and describing things like DCs and specific skill names are simultaneously wordy and specific. I also generally believe that difficulty ratings should come as much from situation as subject: a fence might be easy to climb without pressure, moderately difficult to climb stealthily, and hard to climb in the seconds before the guard dogs reach you. My solution is to simply say "Those within 15' save or die." and let the GM choose the DC and relevant stat in cases where the difficulty is moderate. I may say "save with disadvantage/advantage" if I feel that a description needs a difficulty modifier, and let the GM decide if that means a DC number adjustment or an adjustment to the dice rolled based on their system.

Rule 11: Statblocks go under the key in italics.
Monster statblocks are inherently a mess of symbols that do not flow with the rest of a paragraph. They're also very useful to have on-hand, and are important enough that they're useful to put in a way that's easy to skim for. For each monster in a key, put its statblock immediately after the paragraphs end. Include even non-combatant NPCs and monsters with statblocks described elsewhere in the book.
I also have started including three verbs or adjectives at the end of my statblocks, riffing off of Torchbearer instincts. These are intentionally flexible, meant to give the GM inspirational fallbacks for how the NPC responds to player actions. I find that my NPCs can easily start to act similar to each other if I don't give myself nudges like this, and I find that three words hits a good balance of giving the GM a mental image of who a NPC is without micromanaging the way a scene plays out or writing a bunch of guidance for social edge case scenarios. I generally try to make these different for each key, even for the same monster type, out of a mix of trying to give the NPCs some hint of inner life/personhood and to try to keep repeated encounters distinct from each other.

Rule 12: Number keys in order of escalating revelation, with questions before answers.
I believe in the primacy of a module as a play aide over a book for reading, but a GM's initial reading of a book has direct relevance to the quality with which they run it--that read-through should mirror the process of play, where the most easily reached part of the location raise questions that the reader/explorer must delve deeper to answer. A GM who is engaged by the early part of the module, kept in suspense, and who must read on in order to discover the truth is more likely to retain what they read for use at the table--and more likely to be motivated to bring it to the table in the first place.
When keying a dungeon, it's obviously best practice to have adjacent rooms have adjacent numbers whenever possible. In practice, this rule mostly just means that when you have to choose which branch of the dungeon to describe first, try to get to your most interesting and consequential keys last, and try to finish your levels/etc with foreshadowing of later elements of the book. Don't twist your dungeon map into a mess trying to do this, though--the space itself still takes precedence over the methods used to describe it.
An implicit part of this rule is that modules should contain questions that are revealed over the course of play. This might be full blown mysteries, but it can also be as simple as "what's the deal with all these goblins who moved into this cave?"--whenever possible place the keys full of goblin graffiti before the goblin camp itself.

Rule 13: Books start with common knowledge/rumors/hooks, quickly progress to keys, and save advice/summaries/references/etc for the appendixes.
As an extension of Rule 12, you should not explain all the mysteries of your module in the introduction. A GM reading the book from cover to cover should learn about the setting in loosely the same order that the players will, meaning that pre-key writing should be kept brief.
There are some types of information that are needed before keys, though--the types of information that draw the players to the location in the first place, or information they can gather before arriving. A good introduction serves as a hook, providing a reason for players to send their PCs to the location and GMs a reason to keep reading.
I plan on accomplishing this by framing introductions as common knowledge about the module's subject--less of a god's eye view of the situation and more a set of rumors that the players might be exposed to before they set out. This still follows a Landmark, Hidden, Secret format--the introduction starts with a hook usable by the GM to solicit the PCs into investigating, followed by information they can attain via trivial research, followed by information that may conditionally be gained only by getting the right people to talk, if any.
Anything goes for the appendix. At that point a cover-to-cover GM has read everything in the keys, so spoiler-heavy monster descriptions, faction summaries, and so on become permissible. This should all still be kept to a minimum, but certain sizes and formats of adventures necessitate certain types of GM assistances.

Rule 14: Expect the GM to have the active key in front of them, 20 or so keys fresh in their mind, and up to 80 or so keys already read.
This rule is somewhat intangible, but when deciding what information is worth including or omitting you should assume that the GM has read and reread/prepped the current key and anything else in its section of the book and has probably skimmed nearby sections.
If your book is 20 or less keys, you can assume they have full access to the contents of the keys in their mind; any questions raised and answered in that segment can be seamlessly adjudicated without being a drag on the game. You can drop hints in early keys that are recontextualized by later keys and assume the GM has re-read those early keys with their full context in mind. A Perfect Wife is a great example of a module that works this way.
If the book is 80 or less keys, you can assume the GM read all of it but may not have thought too deeply or prepped for elements in other segments of the module. They show up to session one with full command of level 1 in their mind, but the odds of them not picking up on the implications of setting details that are only fully unpacked on level 3 are relatively high.
If your book is over 100 keys, you can't really assume the player has read (or meaningfully retained) the entire book when they first sit down to play. Some fundamental changes may need to be made regarding book layout to allow the GM to reference commonly known information from parts of the dungeon they haven't read yet. Arden Vul's faction write-ups are an example of what this might look like.

Now that we've listed the rules, let's see how they work in practice! I have here screenshots from a dungeon I wrote for a (possibly?) defunct APA. Click on the sample image for the full PDF.


It's 3 levels, 46 keys, and 15 statblocks in a two-page spread. I've run it a few times to good results, so it seems like a perfect candidate for unpacking into fully written keys. Click the image below for a link to the module rewrite on itch.io.

I've made an effectively unlimited number of community copies available; I hope to use it as a free sample to a larger set of adventures I plan on putting out as part of a project I hope to announce soon.

Okay, so, how did writing under these rules teach work out in practice? I suggest reading the module before continuing, to preserve the effects of Rule 12.

I was extremely happy with the results. Initially I was worried about justifying the extra page count; writing the module as a two page spread forced me to be extremely economical with my word choice in a way that I thought worked well, so if I was going to be unpacking this from two to fifteen pages I needed to make sure that the extra words actually added something; I needed to make sure I wasn't losing my Conceptual Density.

Upon actually writing it, though, I found that the rules I was setting for myself did a great job of avoiding cruft. Grounding everything in an initial paragraph of sensory impressions forced me to think about the spaces I was describing as more than just a set of bullet-point-esque outline entries, and putting extra details in their own paragraphs was good for making me ask myself "does this deserve its own paragraph or can I just leave it unsaid?"

The most fruitful thing I did, though, was completely avoid statements of being in favor of statements of action (Rule 7). As the dungeon's designer it's very tempting to just say 'there is a hidden door in the room' or 'john is a potion merchant', because that's how I think of them when thinking them up, but forcing myself to always describe them as doing something really made sure I had to think about how the players would be interacting with them, not thinking of them as some sort of abstract platonic object--'Pushing on the west wall reveals a rotating panel offering access to a staircase' is much stronger than 'there is a hidden door' in both function and form.

Writing things out in full sentences in this was also gave me a lot of 'hooks' to hang further writing on. Giving every object a verb and framing every personality trait as an action taken forced me to think up a lot of additional imagery and detail. I could see that actually being a net negative if it was done carelessly--if you just fill in the most cliche and expected answer in each of these slots you aren't really adding any new info--but by treating it as a chance to think more deeply about the dungeon as a whole and evoke the dungeon's deeper themes it really helped me think deeply about the dungeon's design; if I'm going to have something worth hanging on those hooks, I need to have a coherent vision for the dungeon's vibes and themes that I can point towards with any evocative language I add.

Those themes don't even really need to be too profound--for this I just wanted to make sure that the bandits felt like they were intruders (lots of damaged high quality furniture), the slimes were out of control and having a great time, the constructs were put-upon and slightly delusional, the villagers were actively absent, and the wizard was mostly concerned with giving inanimate objects life. Whenever I had a place I needed to add a new detail to ground a space's physical description or a verb to illustrate what a NPC is up to, I tried to lean into one of those or to foreshadow some further danger or opportunity deeper in the dungeon. I'm not even sure the slime party fully comes through in the writing, but it was still helpful for me in orienting myself while writing.

I hadn't read Writing Rooms in Pairs when I first wrote this dungeon, but it's something I was loosely trying to do already that I really doubled down on with this rewrite. You should be able to look at any room in the dungeon and see at least one reference to it elsewhere, even if oblique.

The two parts of writing this I most struggled with were perspective (Rule 8) and mechanics (Rule 10). Describing a sensory impression of a room without any assumptions as to the entry point of the players is inherently self-contradictory, and trying to do so without assuming the players know which way is north makes it even tougher. In practice, I mostly either omitted door descriptions (anything I could have written would need to be remixed based on the entry door anyway) or used compass directions (even though those will need to be stripped out by the GM). I gave up on avoiding perspective in a few places--there's a room with one door and a throne at the other side, so I just described the throne as being on the 'far end' because it's hard to imagine a situation where the players enter from behind the throne. It still feels like there must be a better way, though.

The rules for slime infection were a struggle to integrate. I don't want to put them in the intro, because they give away too much about what's going on, but they're a semi-complex set of mechanics that repeat themselves across a half dozen keys. I ended up just putting them on the first page of the dungeon level where they become common, then making sure that the first few keys describing the monsters that use the mechanic make reference to the fact that they attempt to body-snatch humans they overpower. It feels a little inelegant, but it balances spoilers and accessibility so I'm keeping it.

Spells are tricky. They require a ton of words to describe mechanically, change how they work depending on the system, and it's not unusual for a character to know multiple of them. I used common spell names and trusted the GM to be able to connect those names to spells in their own system, with minor name tweaks existing to modify them slightly--Slime Web describes a spell that acts like Web but that's made of slime instead of spiderwebs. I have some misgivings about this approach but don't see a better option yet.

One thing I definitely noticed, though, is that keeping all of these rules in mind as I wrote was very hard. This set of guidelines works better as an editing tool than a writing one--I had to go back and do a few edit passes with a few rules in mind each time to fully apply them all. The longer I write this way the more my first drafts will probably hold to these patterns, I hope.

Finally, deciding what information to include and what simply to imply is always a fun challenge. This module packs a whole bunch of mysteries into a pretty tight space, and I think I did a good job of making all of those solvable without being explicit about them; the three-pack of Elixers of Youth with only two vials left explain why Garl is 15 without me ever saying so explicitly. There are some mysteries that aren't fully answered--why the wizard left--but given that there's no good way of discovering that within the module I think that's okay. A GM could always have the fecal sage give an answer, if they really want, but the important part is just that he's not there any more. I think I would enjoy this level of mystery in a module I read, but I am curious to hear if other feel the same.

Overall, I'm very happy with these rules so far, am going to keep using them on future projects, and encourage other people to think about which of these rules could strengthen your own personal style guides if you started following them. Of course, it's always impossible for the writer to know how their work will look to fresh eyes, so I'd love to hear people's thoughts on the rules in general and the module's style in particular--and if you run the module I'd especially love to hear how it went and if you found the layout easy to reference during play.

This (presumably) concludes my deep-dive into keys and key-writing. You can find the previous posts here:

Part 1: Key Contents

Part 2: Writing vs Prep vs Improv

Part 3: Actual Keys

Tuesday, March 24, 2026

Elements of a Key 3: Actual Keys

 Previously we talked about the types of information that a key or GM provide, and then the types of info best provided in a written key, prepped by the GM, or improvised during play. That was all a little abstract, though, so let's take a look at existing modules and see what information they choose to include or omit.

We'll also start to look at how that information gets presented to the GM. There's been some fun discourse in recent days over if plain text is superior to formatted text, as well as some attempts to standardize NPC statblocks in non-numeric terms, so I figure it's a good time to take a close look at actual examples of module writing as see how different types of information get presented and how we feel about them. I've tried to pick examples that are memorable without being especially spoilery. I'm crediting writers since this is mostly a look at writing, although I'm sure an equally long post could be made about the relationship between keys and art.

I encourage readers to read the keys and take a moment to imagine running them with your regular group before moving on. How would you first describe this key as the players enter it? What questions would they likely ask you through the course of play? How would you answer those questions?


The Dark of Hot Spring Island by Jacob Hurst, Evan Peterson, Donnie Garcia

My interest in key design really started with my initial reading of Hot Spring Island--the sheer aggression of the formatting was an eye-opening moment for me. I've drifted away from the style in the years since, but I still want to start this process with a key from it.

The majority of the key is made up of a list of the objects in the room, with a brief bolded description being paired with additional details within brackets. I'd use the bolded words to describe the players' immediate sensory impression, and then draw on the details as the players start to interact with individual aspects of the room. The key's title summarizes the nook's appearance.

The second paragraph is a plain text description of the more interactive element of the room--the nereid trapped in the painting. It mostly describes her emotional state and history, without many concrete details about how to portray her in the moment, but her faction gets a much longer write-up describing her motivations elsewhere in the book.

The content of the key is mostly occupied with establishing a visual identity for the elves who used to live within the dungeon. The character description is something between lore and an implicit challenge to overcome. It's not especially actionable, although in fairness the module assumes you're adding monsters and treasure yourself; the crystal implicitly creates a great location for treasure to be placed.

A big advantage of this style is the density it allows; the page I copied it from had 10 keys and a detailed map. The bolding of the object names makes it fairly easy to navigate, and follows the natural flow of play (an initial quick description followed by details as players as questions). It's terse but unpackable, with any concepts that can't be summarized in 3-5 words getting their own paragraphs.

It does suffer a fair bit on initial reading, however, and the way that the objects are listed does not follow how speech works--a GM does not typically say "there is a wall, a desk, vases, books, sofas, and a painting in the room". The relationships of the objects in the room aren't clear from the description, requiring the GM to spend time imagining what the room actually looks like. If they've done so as prep this could be a smooth experience in play, but it does still make reading the book for the first time feel very choppy and a little annoying.

The nereid description leans heavy on backstory and generalities. There is no support on her initial reaction to being freed from the painting, but the faction entry on nereids later in the book more than makes up for that.



Holy Mountain Shaker by Luka Rejec

The HSI style has evolved into a more modern general OSE style. Holy Mountain Shaker isn't entirely standard within that style--it describes large locational nodes rather than rooms--but it showcases some of its design tech.

Similar to HSI, the key begins with a set of bolded descriptions paired with parenthetical details and followed by paragraph-based descriptions of concepts too complicated to summarize--in this case listed as bullet points. It follows a loose flow of bolded initial impressions to parenthetical details uncoverable through additional looking to details only discoverable via action below.

As much as the format looks similar, the type of information presented is wildly different. Those bolded descriptions up top are no longer a list of what you see as you first engage with the key, but are instead a list of sub-regions. The HSI style already struggles with items it describes feeling disjointed, but this feels much worse--I really struggle to visualize how these elements relate to each other or how to describe the players initial sensory impressions upon entering this key; descriptions are more evocative (fishy) than concrete. It can also come off as a bit sprawling--it's far less terse than HSI, mostly as a result of it covering much more ground.

The paragraphs serve both as additional details to the bolded items above and as explicit 'if/then' reactions to player actions. It frames the world in how it responds to the players first and foremost, with extra emphasis spent on how long elements take to interact with (this makes sense, the module is on a clock). This seems helpful to me when players perform the actions accounted for, but annoying when the players go 'off the rails'. There's good support for the most likely actions the players might take, but I feel like I'd struggle with all the moments where players poke around and ask questions.

Descriptions also contain significantly more 'tech'--roll tables within paragraphs, additional parentheticals, etc. Sometimes when a player takes an action in response to bullet point information it's accounted for in its own bullet point (search lesser halls->decipher glyphs), but sometimes it's handled within the bullet point itself (touching the mummy).

For such a large space, there is not a lot of 'general information' provided. It specifies that it contains 'bas-relief illustrated scenes', but does not specify what of, and does not give much context as to the history of the space for the GM to draw from while improvising. Concrete and immediate details are better than lore and history when describing a space, but lacking either can create a space that feels empty and non-interactive.


Lair of the Lamb by Arnold K

A series of paragraphs, each of which describes one element of the room. Once each element is described, the more complicated elements of the room are described in more detail.

This has a fair bit in common with the above keys in form. A list of bolded words give you an initial sensory impression, paired with quick descriptions to be drawn for when the players look more closely, and paragraphs below for when they start touching things.

Descriptions follow a flow of starting with a sensory impression, then following that up with a description of what it "is". The first bit gives immediate help to the GM in describing the room, the second gives support in adjudicating player experimentation. The content leans concrete, with lots of explicit clues and a few limitations (non-forceable door) listed to facilitate running the room as a puzzle.

Some reference to other keys is made, including a specific key reference. This information serves as a reminder, and is not actually relevant to this key (there is no way within this key to be told about the fountain).

Some mechanics are described, but are generally described in natural language and are not emphasized. There is a fair bit of leeway in how to mechanically adjudicate setting off the room's trap--are characters within the room caught automatically? Do they have time to run out upon hearing the noise? Does the spray hit characters hiding within the doorframe?

I really like this setup! It's easy to read and easy to reference--following clear rules in its layout that the reader can learn without sacrificing readability. There is some minor formatting (bold text, two bullet points), but it is the exception rather than the rule. Bolding the phrase that best summarizes the paragraph and nothing else feels much more useful for quick referencing than bolding every keyword.



A Perfect Wife by Zedeck Siew

Paragraphs of text describe a location, but then quickly transition to describing the NPC who resides within. The NPC starts with a terse statblock, then a list of 'skills', then a list of things he might do or say. A picture of the NPC sits next to the text.

The paragraphs are non-mechanical other than page references and some bolding, but are extremely economical in their word usage to the point of becoming sentence fragments. The paragraphs are also, above all, concrete but evocative; every description feels like it cuts to the heart of what it's describing with details that the GM can directly repeat to the players while also supporting improvisation. It's also gripping to read as a GM engaged in prep; I have to pause and think for a second after each line, to really let what's written sink in--not because it's confusing, but because it's compelling and rich-feeling.

This is by far the best written key we'll be covering in this post, both from a literary and an information presentation perspective. It provides a mix of concrete initial sensory impressions and examples of NPC action that respond to the most likely questions of the players while also implying a lot of detail about the character, all in an extremely small space.

It does benefit greatly from being a familiar space--unlike the piscine temple above, this is a place where the reader will generally immediately be able to visualize it based on their own prior lived experiences.

I also love the list of 'skills' the NPC has. It's three words that work as a great fall-back if the GM ever needs to decide how the NPC will respond to a situation or how they might treat the players. I wish more NPC statblocks included sections like this.

The bolding can feel a little random, though. It doesn't really correspond to information that would otherwise be hard to browse for.



Deep Carbon Observatory by Patrick Stuart

Paragraphs describe a situation in plain text, preceded by a statblock.

This is notable for being almost nothing but sensory information, presented in order of how visible it would be as the players approach. There is no description of how it responds to the players, what will happen if the players ignore it, or explicit NPC motivations or backstory.

Nevertheless, it provides everything the GM needs to easily prep or improvise most of the above. It provides a great unanswered question (how do you overcome 100 crabs?) and leaves it up to the players to answer. This would require decent improvisational adjudication skills from the GM, but honestly this is the type of problem that's really fun for a GM to solve. It does a great job of making the reader excited to put it into action and see how it unfolds.

It mostly accomplishes this through good scenario design and good writing. The concept is inherently fun, and the writing does a great job of slowly escalating the tension of the scene as the GM reads through it (and by extension as the players approach it physically).



Silent Titans by Patrick Stuart

Bullet-points sprawl erratically through a description in what appears to be a stream of consciousness. The module that exemplifies bullet point discourse.

The key is broken into three large categories--the statue, the justicars, and the scaffold. Each then is filled with smaller bullet-points that provide an unsorted mix of sensory details, advice, and responses to player actions.

The bullet-points are more of a vibe than an information presentation aide. The lack of structure means that they actively obfuscate clarity, featuring the drawbacks of the HSI style (lack of a visualizable space) with none of its economy of space. I have a very hard time visualizing what's going on in this key--I have absolutely no idea what the Scaffold of Memory even is. It feels like I'd have to design this key from the ground up to run it, using the visuals and loose ideas within the key as inspiration.

For all its chaos, though, I actually like the bullet point style as a sort of boxed text. I can imagine myself reading the description of the statue verbatim to my players, using the indents and line breaks the same way I would use them while reading a poem.



Gradient Descent by Luke Gearing

A name, an icon denoting that the space is industrial sized, a title, a terse plain text description, followed by themed bullet points describing individual room elements. The background color denotes that the room is not lit.

Despite being all bullet points and indents, this key feels extremely organized and readable. The initial description and the outermost layer of bullet points create a quick and easy initial sensory description of the space, while the indents section off information that only becomes relevant as the players engage with specific room elements. Information that is hazardous to the players (that the GM needs to not play fast and loose with) is red and has its own special icon.

The writing contains a bare minimum of information that a GM might need to run the key, generally as a string of sentence fragments. It contains limited mechanical information, but is mostly plain language paired with iconography. The information is not especially from a player perspective, and is instead emphasizing points of danger or leverage within the key--there's a lot of dramatic potential energy within the key but not a lot of "this is what the players will do".

The weakness of this approach is that the 'why' of it is not especially apparent. It lists '50 androids', but it is not apparent from this key or others why they are in the crate or even what type of androids they are--Forgotten Androids (deformed factory rejects)? Infiltrator androids (human-appearing spies)? Just bog standard industrial androids (not really appearing anywhere else in the module)? Who put them there? If the players manage to tackle one before they run away, what are their motivations? The economy of information gives the GM exactly what they probably need, but it's easy to imagine situations where the players take the game in directions it actively gives you nothing for--questions it raises that it does not give you the tools to answer.



Wolves Upon the Coast by Luke Gearing

The first line describes what the players initially see, while the rest describes the larger situation in plaintext paragraph form.

Emphasis is placed on the motivations and default actions of each of the NPCs, along with mundane but necessary details (weapons used, treasure hoarded, etc). It also provided a strong point of uncertainty--what should the players do in response to this situation? It contains a danger, a moral outrage, an opportunity for pillage, and insight into the location's history all at the same time.

The writing is also deeply enjoyable to read. Everything feels very evocative, with a lot of inspiration given to the GM in terms of scene-setting and character motivations without ever being spelled out.

The purity of the text does create some annoyances, however. It refers to 'The Ringfort', which is described many pages later and which it gives no clues in how to find. A page reference, or even just a "to the south-west" would have helped me a lot.

A lot of the context is also implied rather than stated, over the course of many keys. The crosses seem to be from Norse invaders who then converted to Christianity, but the Ringfort is Norse descendants not yet converted to Christianity, so it seems it probably wasn't them that did this? A GM can probably just make a call based on this information and not contradict anything else in the book, but the evocative writing definitely asks GMs to nail down their mental model of the key as active prep rather than passive memorization. I think this might be a good thing?



Arden Vul by Rick Barton

13 long paragraphs and 6 statblocks that contain a huge number of the concepts listen in previous post. I can't even bring myself to break this one down because it's simply too big.

This room played a major and repeated part of 3D6 Down The Line's Arden Vul actual play. It is simply a small part of the level, and demonstrates one facet of a semi-major NPC's personality; the rest of his mindset can be discovered via other keys spread among a total of three levels.

The primary quality of this text is size. Anyone who internalizes all of it is well prepared to run a dynamic and deeply internally consistent megadungeon campaign. But it sure doesn't do much to help you internalize it easily.



Keep on the Borderland by Gary Gygax

An extremely early module, written in classic style. One unbroken paragraph, unorganized.

The description starts out with some quick sensory description, then spends the majority of its wordcount describing the tactics the monsters within will use against player characters, ending with some if/then responses to anticipated player actions.

It's amazing how hard to penetrate this is, despite its relatively short wordcount. The lack of line breaks is exacerbated by the lack of informational ordering--parentheticals mention that one ooze may not be present early on, then the same concept is elaborated on at the paragraph's end. A statblock is stuck dead in the center of the paragraph. Significant wordcount is spent explaining the effect of player characters wearing boots on the monster's mechanical effectiveness. It's a rough read.

However, there is something to be said for the amount of thought it puts into the types of actions the players are likely to take. It is clearly designed, but not in the "do this to proceed" mode so much as the "here's a nasty situation and I'm going to make sure you can't easily negate it" one. Of all the keys analyzed here, it's by far the most interested in supporting the GM in their moment to moment unpacking of what would otherwise be vague challenge.


Who included what?

What elements are more or less commonly provided in these module keys? Let's revisit the list from the first post.

Immediate Sensory Impression: Extremely common, with two main strategies; some modules give plain text that can almost be read as boxed text, while others provide raw data on room elements and ask the GM to describe it in their own words. This makes sense, it's the one thing that will definitely need to be handled by every GM every session of play.

Uncoverable Details: Also extremely common, and usually split off from the immediate impressions through some means of formatting. In some cases that split is implicit (sentence one is immediate, two is uncoverable) but in many it's explicit (parentheticals and bullet points). I found that the implicit ones worked better for me, but that modules like Keep on the Borderlands that didn't split it at all were incredibly hard to read.

Immediate Response to Players: Actually fairly rare! Most of these spaces feel like place that exist independent of the players, with NPC reactions to intruders left to the GM's interpretation. This makes sense--you never know what sort of entrance the players are going to make or how they'll look. I feel like this became much more common with the rise of boxed text in 2e and on trad play.

Conditional Response to Players: Somewhat more common, although again with a range from implicit (Wolves and 'those who would come are torn apart') to explicit (Holy Mountain Shaker and its 'when the players do this. . .' prompts). I like the implicit framing quite a bit more--how the key responds to the players can be used as much to tell about the key in general as to resolve a singular situation.

General Concept: Many keys had titles, some of which revealed information that was not otherwise presented in the key. This felt a little like a crutch, but I did appreciate Lair of the Lamb and how it used names to make other rooms easier to remember/find rather than attempting to summarize them.

Temporary/Indefinite/Possible/Reactive Situation: Almost all of the keys leaned heavily on the Indefinite situation over all others, with the exception of Deep Carbon Observatory, who provided an extremely temporary situation and little else, and Arden Vul, which does a little of all four. This makes sense, given that most of these keys are meant to be revisited potentially multiple times but possible and reactive situations are often not worth the word count.

NPC Presence/Motivation/Instincts/Backstory: This one isn't that fair to compare, since simply not all rooms had NPCs. I will say that A Perfect Wife is the clear champ of this category, though, with all four categories fully provided in just a few short and captivating sentences.

Lore & Backstory: Rare but not entirely absent. The most lore-ful was probably Hot Spring Island, but even that information had some immediate relevance. Non-actionable information was, thankfully, omitted.

Other Keys: Some keys referred to other keys and some didn't. I very much appreciated it when they provided a page number whenever they did. I mostly tried to look at keys that made sense relatively stand-alone, so there's not a lot of this.

Authorial Intent: Absent! Although some of the modules do provide it in other sections. I wonder if I should have included Tomb of the Serpent Kings to get a few examples to show.

Kickers & Quests/Potential Player Actions/Unanswered Questions/Exciting Possibilities: This felt like one of the big differences between the keys I liked and the ones that were more forgettable (although this is already selecting for keys that I remembered being interesting from prior read-throughs). The really fun keys weren't simply things you explore because they're there so much as situations that both demand a response but fail to provide an obvious solution. These elements were generally implicit in the key's concept, though.

Mechanical Details: These were often present, always needed, and usually a liability. Attempts to include mechanics mid-paragraph were almost always a significant hit to legibility. The best keys simply listed their statblocks in their own section or explained mechanics in plain language.

Cool Words: Nobody listed here included box text, but many did include segments that could reasonably be read verbatim (or close to it) to players. This actually felt like a major strength of the just-paragraphs keys over the more structured ones--HSI and OSE style thwart this to a significant degree.

The Script: Almost completely absent, but that's probably because I actively avoid modules that do this.

IRL Props: A few of these came with art that you might reasonably show players (while covering up the text next to it). This is a place that Call of Cthulhu really shines, in my experience.

I have no idea if doing an audit like this was fun for you, but it was extremely helpful to me in helping me identify what I actually do or don't enjoy in the modules I use. Next time I'll take a shot at establishing my own house style for what information to present vs omit and how to structure it.