Ten volumes,
one horizon.
Stryx Labs ships products on a codename ladder. Each name claims a real problem. Each problem is solved before the next is begun. Names are revealed in order; dates only earned by shipping.
Mneme.
"A code-graph that remembers what your AI keeps forgetting."
Visit mneme.stryxlabs.com ↗Persistent code-graph memory layer for AI coding agents. Mneme indexes your project once, then exposes 50 MCP tools your AI host can call: recall, blast radius, call graph, audit suite. Architecture and decisions stay attached to the code that depends on them, across every session, locally, forever.
Symbol-anchored embeddings. Three real symbol resolvers — Rust, TypeScript, Python. Drop-in for Claude Code, Cursor, Codex, any MCP-aware client.
- 50 MCP tools
- 22 storage layers
- 14 graph views
- 3 symbol resolvers
- Native macOS · Linux · Windows
- 100% local-only
- No telemetry
- Apply-with-rollback updater
More products are on the way.
A studio with a long horizon doesn't pre-announce. We name our works when they ship, audit them before they leave the door, and never market a roadmap that could change tomorrow.
Want to know first? [email protected]
Want early access to a volume?
Drop us a line. We talk to people who've thought hard about the problem.