Leases and checkpoints for agent runs
6 Jul 2026 · 6 min read
Leases prevent duplicate work, and checkpoints let agent workflows resume without repeating side effects.
A reliable runner needs to know who owns a step and what has already been committed.
AstraRunner treats this as an execution-platform concern: the run needs durable state, traceable decisions, cost visibility, and enough structure for teams to review what happened after the agent finishes.
Patterns that work
- Lease steps before execution.
- Renew leases during long model or tool calls.
- Checkpoint inputs and outputs before transition.
- Recover expired leases into a safe retry state.
Implementation checklist
- Choose lease durations by step type.
- Use fencing tokens for external writes.
- Make checkpoint writes idempotent.
- Alert on repeated lease expiry.
Watch outs
- Clock drift can cause duplicate ownership.
- Long tool calls need heartbeat behavior.
- Checkpointing after side effects risks duplication.
Related reading
How this fits the Astra stack
AstraRunner owns the operational path for agent work: orchestration, scheduling, approval gates, traces, costs, and role-agent handoffs. AstraMemory owns durable context and retrieval. AstraGenie turns those capabilities into product workflows for teams that want automation without rebuilding the platform layer.