Comparison — Tusk's Vault vs the alternatives

Honest comparison to the tools TTRPG groups actually use today. Every claim here is verified against the current code — citations link to the source file that backs them. Each tool listed is good at what it's built for; this page is about helping you pick the right one for your table, not picking fights.

Arrived from the Foundry module? Lite and the full Vault is the section you want: what the module answers on its own, and what the application adds to it.

🆚 At a glance

Tool	Where it runs	What it costs	How it answers	Where you ask it
Tusk's Vault	Your own machine	Free · MIT	Cited, from your own documents	Discord · Foundry · any MCP client
Tusk's Vault Lite — the Foundry module alone	Your browser, inside Foundry	Free · MIT	Cited, from journal entries in the folders you choose	Foundry's chat bar
NotebookLM (Google)	Google's servers	Free with a Google account	Cited, from what you upload	Its own tab
World Anvil	Their servers	Free tier · paid plans	You search and read it yourself	Its own tab
LegendKeeper	Their servers	Subscription	You search and read it yourself	Its own tab
Kanka.io	Their servers, or yours	Free tier · paid plans	You search your own records	Its own tab
Obsidian + AI plugins	Your own machine	App free · plugins vary	Whatever the plugin does	Its own window
ChatGPT / Claude.ai direct	Their servers	Free tier · paid plans	From whatever you attach — and invents when you haven't	Its own tab
D&D Beyond	Their servers	Free sheets · paid books	Rules and characters, not your campaign	Its own tab

No ticks and crosses here. Running in the cloud is not a defect, and neither is having no AI in it: a wiki that renders a setting beautifully is doing the job it was built for, and a research notebook is meant to be read alone. Half these tools are a better answer than this one for the thing they were built to do, and a grid of red crosses would only have hidden that. The table says what each one is — the sections below say who should pick it.

It is still a table written by the person who made one of the rows, which is what makes a flattering row worth nothing on its own. So, plainly, what this one loses:

Setup. It wants Node, Git and a terminal before it will do anything.
Always available. It answers from the machine it runs on, so a sleeping PC is a silent bot. (You can still ask from a phone via Discord — as long as that machine is awake.)
Corpus size. The folder source stops including documents past roughly 200 pages (known issues).
Support. There is no team behind it and nobody to escalate to.

Discord is Vault's oldest surface, not its only one: the same answer pipeline also serves Foundry VTT (via a companion module, /tusk <question> in the chat bar) and any MCP client — Claude Code, Claude Desktop, anything that speaks the protocol. See Foundry VTT. Session recordings become documents in that same folder by way of Tomes, so what the table above calls "your own documents" grows without you writing it up. "Free" means the software — MIT, no account, no subscription. You still pay whichever model you point it at, or nothing at all on Ollama.

🧩 Lite and the full Vault

The Foundry module runs two ways, chosen in its own settings. Lite answers on its own, out of the journal entries in the folders you choose, with nothing else installed at all. Bridge points the same chat bar at Tusk's Vault running on the GM's machine. Same /tusk command either way, and the same citation rules — what changes is what it is reading, and what it is allowed to do with it.

If you came here from the module, this is the row-by-row version of what downloading Vault actually gets you.

	Lite — the module alone	Tusk's Vault — the application
To install	Nothing beyond the module	A free download, plus Node and Git
Where your lore lives	Journal entries in the folders you choose — so wherever that world is hosted	The notes you already keep — Word, PDF, Markdown, or a whole Obsidian vault
How much it can read	Only what fits into a single question	A folder of documents up to roughly 200 pages, all of it sent with every question; an Obsidian vault can be far larger, because past a budget it sends a map plus the notes each question actually needs
Finding the right note	Matches the words you typed	Works out what you meant, so the page still turns up when you word it differently
Models	Gemini, on your own key	Gemini, ~400 more through OpenRouter, a Claude Code subscription you may already pay for, or a local model for nothing
With no API key at all	Still searches, and quotes the entries that match — nothing leaves your machine	A local model through Ollama answers in full, for nothing
Where your API key sits	In your browser — which every other module you install can read	Encrypted on disk and tied to that machine, so a backup, a sync folder or a support bundle carries nothing usable; never in a browser. Software already running as you can still read it
Voice	One archivist	Eight personas, or one you write
Guessing	Won't, ever	Speculative Mode reasons from established traits, cites what it used, and tags the answer `[speculation]` so it is never mistaken later for something you wrote
When it doesn't know	Tells you, and forgets	Records it as a lore gap; you answer once, and it applies your ruling to the same question asked differently
Who may ask, and who sees it	A per-world ask policy. It can additionally filter by Foundry's own permissions, per asker — but that is opt-in and off by default: out of the box it answers from every entry in the folders it reads	A per-world access mode and a per-install ceiling, but no notion of Foundry's permissions: it answers from the whole archive
Also answers in	—	Discord, and any MCP client — Claude Code, Claude Desktop, anything speaking the protocol
Session recordings	—	Tomes turns them into chronicles, into the folder Vault reads
Price	Free · MIT	Free · MIT

Read the permissions row twice, and then go and set it. Lite can be the safer of the two, but only once you choose it: the module ships with its lore scope set to answer from everything in the folders it reads, and the per-asker mode — where it asks Foundry what this particular player is allowed to open and answers only from that — is opt-in, in the module's own settings. Until you switch it on, a player asking Lite can be told about an entry they cannot open themselves.

Vault has no notion of Foundry's permissions at all: it answers from your whole archive. So on either half, opening it to players is a decision about spoilers you make yourself — see Foundry VTT before switching that on.

Stay on Lite if the campaign already lives in Foundry journals, it fits, and you would rather not run anything on your own machine. It is not a trial, it does not expire, and nothing nags you.

Download Vault when you hit one of the walls above: notes that are not journal entries, a campaign too big for one question, a question worded differently from the note that answers it, or an API key you would rather was not sitting in a browser. Nothing is migrated and nothing is lost — you switch the module from Lite to Bridge in the same settings panel, and your journals stay exactly where they are.

🔍 Vault's claims — verified against the code

Every claim made for Vault above is backed by code that exists in this repo:

Claim	Where it's enforced
Local-first	Default bind is `127.0.0.1` — see `host-origin-guard.ts`, which refuses requests whose `Host` is not a loopback literal.
Source-grounded	Canonical seven-rule block in `prompt/system.ts → CANONICAL_RULES_BLOCK` is force-appended whenever the chosen system prompt doesn't already contain it — citation discipline is structurally guaranteed, not configurable away.
Off-corpus refusal	Rule 3 requires the bot to emit the canonical refusal phrase verbatim; the lore-gap detector picks it up via `LORE_GAP_TRIGGER_FRAGMENT` and routes the question to the dashboard's Lore Gaps tab.
Discord-native	Bot client in `src/server/discord/`, message handling in `src/server/surfaces/discord.ts`.
Three ask surfaces, one pipeline	`SurfaceId` is `discord \| foundry \| mcp` (`chat/types.ts`); all three call the same `chat/ask.ts`, so persona, corpus and citation discipline are identical wherever the question came from.
Encrypted keys	AES-256-GCM, machine-bound, in the platform config dir — see `keys/crypto.ts`. Never plaintext on disk.
MIT open-source	LICENSE.
One outbound network call per answer	The LLM adapter — `openrouter.ts` or `gemini.ts`, Ollama over loopback, or a local Claude Code process that talks to Anthropic on your behalf. Discord adds its gateway connection when that surface is on; Foundry and MCP add none. No telemetry, no analytics, no phone-home.
Nothing to install inside the app	Local models and personas are ordinary features, not modules. There is no plugin system to learn and nothing to enable.

Lite's half is not in this repository — the module is its own codebase, and the rules it holds itself to are restated there. The version-pinned contract covers the Bridge wire protocol, which is the part the two share; Lite never talks to Vault, so its citation and refusal rules are asserted in the module's own suite rather than cross-checked from here.

⚰️ What happens if this project dies

Worth asking of anything you are about to trust with years of worldbuilding, and it is the fair question to put to a project maintained by one person.

Your source material is untouched. Vault reads the folder you pointed it at. It does not convert your documents, import them into a database, or move them — there is no database, and Tusks-Lore/ after a year of use contains the same files you put there. Delete Vault and the folder is exactly what it was.

There is no export step, because nothing was captured. Everything Vault adds sits beside your documents as plain, pretty-printed JSON you can open in any editor:

File	What it holds
`clarifications.json`	The rulings you recorded, question and answer as text
`lore_gaps.json`	Questions your notes did not cover
`clarifications.embeddings.json`	Base64 vectors used for matching — derived data, regenerable

An export button exists in hosted tools because the tool is holding something of yours. Nothing here is holding anything.

Nothing switches off, because nothing is running anywhere else. There are no servers, no account, no licence check and no phone-home, so there is no lever anyone can pull to stop an install that already works. It keeps working whether or not the project does.

The code is MIT. Fork it, patch it, hand it to whoever wants to keep it going, or lift the one part you liked. LICENSE.

What you would actually lose

Not nothing, and the honest version matters more than the reassuring one:

Fixes stop. Bugs on the known issues list stay there.
Nothing adapts around it. If Discord changes its gateway, or a provider changes its API or retires a model, an unmaintained Vault breaks at that surface and stays broken until someone patches it. This is the realistic way it dies — not deletion, but drift.
Dependencies rot. The security advisories that ship in any Node project stop being triaged.

So the risk is real, but it is bounded, and it is a different shape from the one hosted tools carry. There, the failure mode is your material becoming inaccessible on somebody else's schedule. Here, the failure mode is a tool that stops improving next to a folder of files that never depended on it.

🧭 Tool-by-tool — the nuance

One caveat before the list: everything said about Vault below is checked against this repository's code, but everything said about the other tools is a snapshot of products that are actively developed and priced by other people. Several of them are adding AI features. Check the current state of anything you are about to pay for rather than taking this page's word for it.

Lite is not in this list — it is the same project rather than an alternative to it, and it has its own section above.

NotebookLM (Google)

This is the closest comparison, so it gets the longest entry — and the concession first.

What it's great at, and better at than this: cloud-hosted retrieval over documents you upload, with exact source citations and a genuinely excellent interface. Free with a Google account, nothing to install, and built by a large team with a budget — it will keep improving faster than a one-person project can.

If what you want is to interrogate a pile of documents carefully and on your own, use NotebookLM. This is not trying to beat it at that.

What it does not attempt. The distinction worth drawing is not quality, it is purpose. NotebookLM is built for a reader alone with their sources. Almost everything below is something a research notebook has no reason to have an opinion about:

It is a tab you visit, not a place answers appear. Vault answers in your campaign Discord and in Foundry's chat bar, where the argument is already happening and everyone at the table sees the reply. That is the difference between settling a question and looking one up. (A per-world access mode decides whether players may ask at all, and whether answers are whispered or posted — see Foundry VTT.)
It has one voice. Vault ships eight personas, from a scholarly archivist to a foul-mouthed sellsword, and the citation rules survive all of them. A table's tone is half of why anyone enjoys asking.
It will not speculate, deliberately. "Who is most likely to start a tavern fight" has no answer in anybody's notes. Vault's Speculative Mode reasons from established character traits, cites the evidence it used, and tags the conclusion [speculation] so a guess is never mistaken later for something the GM wrote. NotebookLM's grounding is a design commitment, and a correct one for research; it just rules out the questions a table most enjoys.
It sanitises grim material. Campaign fiction inherits its genre — war, cruelty, horror, grief. Vault's prompt states plainly what the corpus is (authored fiction, the asker is one of its authors, the task is recitation) and recounts a scene as written. A summariser that softens the author's own writing back to them is failing at the job.
A GM ruling is not a first-class thing. Vault's clarifications persist, embed, and get re-applied to differently-worded questions later, outranking the documents when they disagree. NotebookLM has no concept of "the GM overrides the source".
Your documents leave your machine. Uploads live on Google's infrastructure. Vault keeps the corpus and the index local, sends only matched passages to the model you chose, and can run entirely offline against a local model.
You cannot change it. Vault is MIT, forkable, and lets you pick the model — or swap it mid-campaign.

Pick NotebookLM if you want a polished cloud research notebook, you are mostly working alone, and you are content for Google to hold your campaign.

Pick Vault if the answer needs to land where your group can see it, in a voice they picked, about material a cloud summariser would flinch at — and you would rather the whole thing ran on your own machine.

Honestly: both. They are not substitutes. Nothing stops you researching a setting in NotebookLM and running Vault at the table.

World Anvil

What it's great at: worldbuilding as a hypertext wiki. Beautiful presentation. Excellent for setting publication.