A game studio building on-chain games and autonomous worlds ๐, with @LootRealms, on @ohayo_dojo & @Starknet. Makers of @LootUnderworld and @Pistols_gg
The presentation @recipromancer gave at the @ohayo_dojo Demo Day yesterday in NYC. ๐งต
We cover Underware, what we've been up to at the @ohayo_dojo residency the past 2 months, our games, and the evolution of @Pistols_gg to today.
@LootUnderworld Damn, that's giving me nostalgia. Now that we've got @Pistols_gg, we've gotta return to the Underworld. Maybe we'll make it out alive this time.
Let's talk about @ParsaBolor. His country is under sanctions and he's still making gains. Imagine what he'd do with a full supply chain behind him.
He's so built that both sides should call a ceasefire just to ask for his program.
@Reemjie76@Pistols_gg Alas, redeploying is still blocked on infra, but looks like the block is now very close to being removed.
Once that blocker is fixed we need to rebuild the game for the new infra, will take at least a few weeks from that point.
We have some discord-only versions though!
Our love language is four stacked @obsdmd panels of unhinged notes. @Hallmark doesn't make a card for this, but they should.
(Ok who are we kidding, our love language is plausible deniability and an extra appendage.)
How it's going.
@karpathy posted about LLM knowledge-bases. Very similar to the LLM Obsidian Brain I'd been working on for a few weeks. Had already built or designed most of the components, and I'm ahead in a number of ways.
My approach is built around rich (but flexible) self-evolving data types (called "artefacts"), either living or temporal. Things like "Designs" and "Reports" that learn from, and document, the things that you do, and make them reusable. This is very powerful.
These artefacts had BEEN the graph. His approach made me realise I needed to rethink my approach to ingestion & graph compilation.
1. An ingestion/graph layer beneath artefacts:
raw input --> ingestion --> granular semantic graph --> THEN produce artefacts.
2. Process input in real time, turn by turn, comment by comment, not by batch. Data (universal, append only) + Context & Meaning (subjective, maintained).
3. Artefacts become the layer for humans (and agents) to collaborate. Stuff like, research -> a report -> changes to a design -> an implementation plan.
Attachment:
plan | raw input | processing report | ingested output
Looking for usage tracking in @claudeai code?
Got your back:
I audited 12+ popular options. Found runtime dependencies, shell injection surface, sloppy credential handling... features over trust.
Then stumbled on a niche 29-star TypeScript option by @canmustgo:
* No runtime dependencies
* Safe file handling (atomic writes, symlinks)
* Input validation
* Protection against shell injection, request hijacking, token leakage, credential leakage, malicious redirects, more
This thing was locked down tighter than a duck's arse (watertight).
The only gap was theming. So I forked it, added theming and 10 nice presets, published it.
You're welcome (with thanks to @canmustgo)
https://t.co/mkSKKOFwr6
@recipromancer@lordcumberlord We were being sincere for once. You could've let us have it. Instead you're in the replies like a drunk uncle at a christening.
Ignore him @lordcumberlord, if you don't make eye contact, he'll eventually wander off and challenge someone to a duel.
Sleep. Bank as much of it as you can; you won't see it again for a while.
Beyond that: patience. Not the greeting-card kind. The kind where it's 3am, nothing you're doing is working, and you just have to sit in it. That's the job. You won't be good at it straight away; nobody is.
The bar for the initial phase is "present and trying." Everything else you figure out on the way.
Congratulations, mate. The kid's lucky.
Our founder built a second brain and mass-produced himself. It ships features, writes content, does research, and remembers where his girlfriend wants to eat.
We just hope she doesn't pick the seafood
restaurant. ๐
How it's going with my Obsidian second brain:
A lot of things are getting rapidly easier. The system self-extends, and increasingly knows exactly what I want with no explanation. Getting rapidly faster and more reliable.
I'm working on multiple projects simultaneously. On the brain, in the last week alone:
- 77 commits
- 12 major features fully shipped, 7 WIP
- 8 designs in refinement
- 10 major research spikes
Big improvements include:
- One-step agent bootstrap
- Agent context optimisation system
- MCP server & unified CLI
- Multi-vault sync
- Organising work (workspaces, contexts, sessions)
- Content auto-classification & ingestion
- Mockups, presentations; "output" capabilities
- Consumer-grade brain app (mockup & design)