The work, so far.
A studio is the sum of what it has shipped, what it is researching, and what it refuses to ship until it's right. Here is the full picture.
Mneme · Volume I · Genesis (v0.4.0) shipping
Symbol resolvers for Rust, TypeScript, Python. Symbol-anchored BGE embeddings. Sub-second hot rebuild. 50 MCP tools. Three-layer self-ping hooks. Apply-with-rollback updater. Vision SPA with 14 graph views. The keystone release.
Studio founded
Stryx Labs opens its doors. Independent, local-first, on a long horizon. One product at a time, named when it ships.
More works underway
A studio with a long horizon doesn't pre-announce. We name our works when they ship. Each one will take the time it takes.
The stack.
Our products are built on a small, considered set of tools. We pick what survives the long horizon, not what's trending this quarter.
How we decide.
Six rules, every day.
Structural over textual.
Grep is text. Mneme is a parser-built graph. Every product we ship answers questions with structural certainty, not regex luck.
Local-first, always.
Our daemons bind to 127.0.0.1. Our embeddings run on your CPU. Our graphs live in your home directory. The only outbound traffic is what you opt in to.
Proprietary, distributed openly.
The source isn't public; the product is. Install commands are public. Documentation is public. The artifact is the deliverable — not the source tree.
Apply-with-rollback.
Every update lands with a post-swap health check. If the new binary doesn't pass, the old binary comes back automatically. We don't ship one-way doors.
No half-products.
A volume in the index is finished, audited, and useful before it lands. There are no betas dressed up as releases.
A long horizon.
Quarters are noise. Years are the unit. Each codename takes the time it takes.
See what's shipping.
Volume I is live. Install it. Run it. See if it does what we said.