Directed acyclic graphs orchestrating multi-agent workflows with dependency resolution, parallel execution and checkpoint-based resumption.
Scheduling
Units are ordered topologically, then anything independent is released together. Parallelism falls out of the dependency structure rather than being guessed at.
Isolation
Before a wave opens, the planner checks file overlap across its units. Two agents are never handed the same file, so parallelism costs nothing in merge conflicts.
Recovery
Each node checkpoints on completion. When one fails, the graph replays from that node forward and everything already proven stays proven.
Dependency-aware scheduling with topological ordering.
Independent subgraphs run concurrently for throughput.
Retry from the last good node, never from the top.