A Call of Cthulhu campaign stalls on a lost clue, not a lost fight
The honest version, same as our Daggerheart and Pathfinder pages: this product was built around D&D and two of its prompts say so in plain text. None of it is rules-aware, so a Call of Cthulhu session goes through the same pipeline and comes out the same shape. What is different here is what the record is for — in an investigation game a transcript is an evidence log, and that changes which of these tools earns its keep.
The problem
An investigation stalls differently from a dungeon crawl. Miss a plot beat in a fight-driven game and the campaign routes around it; miss a clue in Call of Cthulhu and the table grinds to a halt three sessions later, with nobody sure what the letter actually said, which name the librarian gave them, or whether anyone ever followed up the second address. The Keeper is then choosing between re-feeding the clue and watching the table flounder, and both cost the scenario something.
Nothing in the pipeline is rules-aware
Worth stating exactly, as on every system page in this cluster. Transcription turns speech into text; extraction pulls out the people, places, items and organisations that were mentioned; the recap summarises what was said. No step reads a rulebook, tracks a statistic, or cares what dice you rolled. What happened mechanically at your table is invisible here.
So there is no Call of Cthulhu support to switch on, and equally nothing that breaks. Your sessions transcribe, the NPCs and towns and cults land in the wiki, the recap covers what happened, and you can ask questions about your own campaign afterwards — exactly as for D&D, Daggerheart, Pathfinder or homebrew. Extraction categorises what it finds dynamically, from context, rather than against a fixed D&D list, which is why a 1920s campaign of investigators, newspapers and occult societies does not come back crammed into D&D-shaped buckets.
What the product does not do is also unchanged by system: no character sheets, no dice, no maps, no handout hosting, and no speaker labelling — a transcript does not say who spoke.
Two prompts say "D&D", and what that costs a 1920s game
In two places the AI is told, in so many words, that it is looking at a D&D game: the transcription prompt opens "A Dungeons and Dragons tabletop session", and entity extraction asks for entities from a "D&D session transcription" before categorising what it finds from context. We would rather show you that wording than have you find it later.
For Call of Cthulhu the prefix probably costs a little more than it does a fantasy game. It primes the model to expect fantasy proper nouns, and half of a Cthulhu session is deliberately mundane — surnames, street names, newspapers, shipping lines — while the other half is vocabulary that was never meant to be said comfortably aloud. A speech model has workable priors for an invented elf; it has none for Nyarlathotep said quickly by a nervous player. The fix is the same one the other system pages describe, and it matters more here.
The glossary is where the Mythos names survive
Each campaign has a glossary you control, fed into transcription ahead of every session. Put the scenario’s proper nouns in it — the investigators, the town, the family names, the cult, the tome — and they start coming back spelled consistently instead of phonetically.
It takes pronunciation hints too, rendered like "Wendelbrie (WEN-del-bree)", and that field earns more here than anywhere else on this site: a hint about how a name sounds is worth far more to a speech model than a spelling it could never guess from sound. There is a budget — the transcription API honours about the last 224 tokens of a prompt, so the glossary is budgeted in tokens (a few hundred characters of names in practice, 850 at the absolute most) — so it is a list of your most-confused names, not the whole Mythos. The Whisper guide has the underlying detail.
A transcript is an evidence log
This is the argument for recording an investigation game specifically. A dungeon crawl survives fuzzy memory because the next fight happens either way. An investigation does not: the campaign is the chain of clues, and every one of them entered play as words said aloud at the table. A transcript preserves exactly that layer — every handout as it was actually read out, every NPC interview verbatim, checkable months later. Not what your prep notes meant the clue to say; what the players actually heard.
"What did the librarian actually tell us" stops being a memory contest. Ask Your Campaign answers that kind of question from your own sessions, with a citation to the session it came from — which matters in a genre where the difference between "a black book" and "a black ledger" can decide where the party goes next. The wiki meanwhile accumulates every named person and place with the session they first appeared in: a running cast list for a genre that keeps killing its cast.
It also settles the question Keepers rarely say out loud: did the clue ever leave your prep notes? Every Keeper has watched a table miss something obvious and wondered whether it was missed or never actually delivered. The transcript answers that, which tells you whether to re-feed the lead through another channel or trust the table to get there.
Handouts, journals and typed notes come in as text
Cthulhu scenarios run on props and paper — letters, clippings, diary pages — and a handout passed around the table in silence is invisible to a recording. The ones that carry clues are worth reading aloud anyway; whatever is read aloud lands in the transcript like everything else said at the table.
For the rest there is transcript import: if your table keeps a typed session log, runs interludes or downtime in text, or wrote up its early sessions before anyone recorded, paste the text in and it runs the same chain — entities into the wiki, a recap, the lot — while consuming no transcription hours. Every plan includes imports: two a month free, ten on Dungeon Master, unlimited on Guild Master.
The numbers, with sources
| What | Value | Source |
|---|---|---|
| Call of Cthulhu — how Chaosium describes it | "Call of Cthulhu is the world’s best game of secrets, mysteries, and horror", run by "the Keeper of Arcane Lore—the Game Master who will present the scenario to the other players" (Keeper Rulebook, 7th Edition) | sourceas of 2026-09-20 |
How we sourced this: The Chaosium row above comes from the publisher’s own store page and covers how they describe the game and its Keeper — nothing more. This page makes no claim about how Call of Cthulhu’s mechanics work, because describing rules from a product page is how confidently wrong pages get written. Everything said about DM Scribe’s own behaviour — the two D&D-worded prompts, dynamic categorisation, the glossary, pronunciation hints and the token-counted prompt budget — was read from this product’s source code.
This is not for you if…
Same test as the rest of this cluster: how your table plays decides it, not which rules it uses.
- Your Call of Cthulhu is one-shots and convention scenarios. A single evening leaves nothing to forget by next week; this earns its keep across a campaign long enough to lose a clue in.
- You want rules-aware tooling — character sheets, dice, tracking Sanity at the table. None of that exists here and none is planned. This handles what was said, not what was rolled.
- You do not record your sessions and do not want to start. Everything here is built on a recording, and your table’s consent to being recorded comes before any of it.
- Your table is play-by-post. Every plan imports text transcripts (two a month free, ten on Dungeon Master, unlimited on Guild Master), but a text-first campaign is not what this is shaped around.
Other ways people use this
- Actual play and podcast crews
Actual play crews already have clean multitrack audio. Turn it into show notes, a searchable archive and accessibility transcripts.
- Professional and paid GMs
Running paid games for several tables is the worst continuity problem in the hobby — and a business reason to solve it. What keeps four campaigns straight?
- West Marches and open-table campaigns
Twenty players, rotating parties, one shared world — and nobody who was at every session. A shared record is the format’s missing piece.
- Daggerheart tables
Nothing in the pipeline is rules-aware, so Daggerheart transcribes and extracts like any other game. The prompts say D&D. Here is what to do about it.
- Pathfinder 2e tables
Pathfinder sessions transcribe, extract and recap like any other game — nothing here is rules-aware. What the D&D-tuned prompts cost, and the fix.