We build the layers around intelligence: persistent agents, durable memory, traceable execution, and the infrastructure that connects actions to outcomes.
Explore the lab ↗ · Meet the systems ↗ · Read the research ↗
Agents · Memory & identity · Execution · Evidence & settlement
A model can answer a question. A system can carry the work forward. We build the connected layers that make that possible — with people in control and outcomes you can inspect.
|
A deterministic execution and accounting network for agent activity, with signed activities, replayable state, and receipts; checkpoints connect execution to Paxeer settlement.
|
A local-first agent workflow engine in one C11 binary: code graph, specs, snapshots, memory, and a durable fleet for getting software work qualified and merged.
|
|
An open identity and provenance protocol for models, agents, harnesses, datasets, and the artifacts connecting them. Signed records form verifiable lineage.
|
A companion AI agent for ongoing work, designed to learn, adapt, and help get tasks done across the Centra workspace.
|
Stars and repository counts above update automatically. Projects may be experimental or pre-release; check each repository for its current status.
| Layer | Systems | What it explores |
|---|---|---|
| Workspaces | Centra · Workmates | Ongoing work with personal agents and AI teammates |
| Agent architectures | Sturgis · Keith | Continuity between calls and safely tested self-improvement |
| Execution & settlement | LayerX · Paxeer | Ordered activity, verifiable receipts, custody and exits |
The Prometheus Papers explore agent memory, coordination, evaluation, and security. Start with Context Windows Are Not Memory or browse the full open-access archive. We build what we study, and working systems give us new questions to investigate.
Build with us. Question the work. Follow the evidence.
Website · All repositories · Research · Get in touch
© 2026 Sidiora Labs · Applied research. Working systems.