02 / Shinraya Tech Labs
Built to investigate.
Engineered to be used.
Open implementations that support research and put ideas to work. Organized by the problems they address, with maturity and evidence kept in view.
Observe & understand
Make the behavior of complex systems inspectable.
AgentWatch
Runtime evidence, provenance, comparison, and replay for AI-native systems.
The v3 README describes a passive instrument; most analysis capabilities remain experimental.
View AgentWatchPulseStack
Observability and execution replay for distributed AI workflows and agents.
The repository identifies the platform as under active development.
View PulseStackBuild & orchestrate
Connect ideas to executable, repeatable workflows.
RepoPilot
A repository agent that edits code, runs tests, and iterates on execution feedback.
Test success is part of the workflow; it is not a guarantee of correctness or safety.
View RepoPilotVEX
Content-hashed incremental builds with task skipping and a shared remote cache.
A Rust implementation. No comparative performance claim is made here.
View VEXFluxCore
Event-driven orchestration for reactive pipelines, services, and automation.
Public implementation and architecture; production scale is not independently verified.
View FluxCoreExplore & apply
Carry experimental methods into working software.
MEOWCODE
A unified AI workspace that routes requests across configured providers.
The README describes a Next.js frontend, Fastify API, and Postgres-backed deployment.
View MEOWCODEMYCELIUM
An evolutionary runtime for generating, mutating, and evaluating executable software.
Architecture and runtime are documented; generalized self-improvement is not established.
View MYCELIUMSentinelCV
Computer vision for fall detection and safety monitoring with CCTV and IP cameras.
A public implementation; field reliability and safety certification are not verified.
View SentinelCVBuilt in the open
Good questions deserve
more perspectives.
Reproduce an experiment. Challenge an assumption. Build something useful with us.