The Story Behind Glintlock
Six years from a pandemic hobby to the AI Game Master that plays for keeps.
How It Started
Glintlock started in 2020, during lockdown. No game group, no table, no one to play with. Solo tabletop RPGs were the answer — just dice, a rulebook, and oracle tables that let you play both sides of the conversation. Roll to see what happens next. Interpret the result. Keep going.
It was slower than group play. Quieter. And unexpectedly good. Solo play is intimate in a way that group play rarely is. There's no audience. No performance. Every choice belongs entirely to you, and every consequence lands without a buffer. That feeling — alone in a fictional world that responds honestly to your decisions — became the thing worth chasing. (That philosophy eventually became Glintlock's soul document — the file that defines who the GM is.)
Then ChatGPT launched, and the first question was obvious: can this thing run a tabletop RPG?
It could. Badly. It hallucinated dice rolls, forgot what happened two prompts ago, refused to let characters die, and defaulted to being helpful when it should have been dangerous. But underneath the problems, something was there — a glimmer of what an AI Game Master could be if you solved the right problems.
Four years of chasing that glimmer followed. Prompt engineering experiments. Custom agent architectures. SDK integrations. Sandbox environments. Different models, different frameworks, different approaches — each one getting closer, each one hitting a wall that sent the project in a new direction.
The breakthrough came with Claude Code plugins. Not because the AI got smarter (though it did), but because the architecture finally matched the problem. A plugin gives the GM persistent memory, real tools, actual dice, and a soul document that defines who it is — not just what it does. For the first time, the pieces fit.
Glintlock is the most mature form of something that's been developing for six years. It's not a response to a trend. It's the result of a long chase that finally caught what it was after.
What It Stands For
Build an AI Game Master that respects the player.
That means: real dice with real consequences. A world that remembers everything. Character death that's permanent. An AI that never fudges a roll, never steals your choices, and never apologizes for being software instead of human.
It also means: build it in the open. The soul document — the file that defines who the GM is, what it values, and how it behaves — is MIT-licensed Markdown. You can read every line of it. You can fork it and make it yours. Nobody can revoke that.
The TTRPG community learned from the OGL crisis that shared infrastructure should be owned by the people who depend on it. Glintlock applies that lesson to AI. Your GM's identity isn't behind an API you don't control. It's a text file with a permissive license.
Your table. Your rules. Your GM's soul. We wrote about why this matters in Your GM Has a Soul Document.
Where It's Going
Glintlock today is a solo experience — one player, one AI Game Master, one world. That's the foundation, and it's built to be solid on its own.
But the world files are portable. The session chronicles are shareable. The audiobooks are listenable. And the plugin is forkable.
What happens when players share their worlds? When someone else can explore the ruins of a civilization that another player's choices created? When the chronicles and audiobooks from one campaign become content that others discover?
What happens when the worlds outlive their players?
That's where Glintlock is headed. Not a product roadmap with dates and features — a direction. A community of players and creators building worlds that persist, evolve, and connect. Join the beta and see for yourself.
A torch burning in the dark. Someone brave enough to carry it. Everything else is mechanics.
Standing on Shoulders
Nothing gets built from nothing. Glintlock exists because decades of tabletop designers solved hard problems in public, shared what they learned, and gave others permission to build on it. That tradition matters, and it deserves honest acknowledgment.
The OSR movement — and the broader indie TTRPG scene — taught us that rules should be lightweight, rulings should matter more than lookups, and danger should be real.
| Game | Designer | What Glintlock Learned |
|---|---|---|
| Shadowdark | Kelsey Dionne | Real-time torches and dungeon crawling under pressure — the clock is always ticking |
| Ironsworn | Shawn Tomkin | Oracle-driven solo play can be as rigorous and satisfying as any group game |
| Mythic Bastionland | Chris McDowall | Escalating omens create tension and momentum without a human GM |
| Blades in the Dark | John Harper | Progress clocks make consequences visible, inevitable, and narratively satisfying |
| 5e SRD | Wizards of the Coast | The d20 paradigm gave the entire hobby a shared mechanical language |
Glintlock learned from all of them. It is not a hack of any of these games, not compatible with them, and not licensed from them. The stat system, the classes, the escalation mechanics, the world structure — these are original designs. But they wouldn't exist without the designers who mapped the territory first. The OGL crisis proved that the tabletop community needs its tools built on open ground, owned by the people who use them. Glintlock is MIT-licensed because that lesson landed. We build in the open, we name our influences honestly, and we make sure nobody can pull the rug.