Recoverable, verifiable workflow memory for AI agents on 0G.

0G OpenClaw Memory Runtime is a persistent memory and verifiable recovery layer for OpenClaw-style agent workflows. The project addresses the issue of agents losing context and being unable to prove the consistency of recovery states after process crashes, service restarts, or runtime migrations. The system receives workflow events from the Go orchestrator, the Rust runtime performs deterministic replay of events and generates checkpoints, uses 0G Storage to persist the checkpoints, and anchors the workflowId, stepIndex, rootHash, and cidHash through MemoryAnchor on the 0G Chain. The project supports capabilities such as hydrate, verify, and trace, enabling a run to not only be recovered but also verified again.
During this hackathon, we completed the OpenClaw-style workflow events ingest, Rust deterministic replay and checkpoint generation, 0G Storage persistence, 0G Chain / MemoryAnchor anchoring, hydrate / verify / trace core links, as well as the dashboard and verify console for the judges. The current project has formed a complete demonstration path from ingest -> checkpoint -> restart -> hydrate -> verify -> trace, and includes README, evidence docs, and judge demo docs.
Pre-funding