Agently TriggerFlow State And Resources

This skill covers TriggerFlow runtime state placement and runtime-only dependency design in Agently. It focuses on runtime_data versus flow_data, flow-level versus execution-level runtime resources, execution-scoped data.set_resource(...), what survives execution save/load, what does not, and how state or resources cross sub-flow boundaries. It does not cover generic branching patterns, model requests, interrupts as a standalone topic, flow-config export/import, or the mechanics of execution restore.

Prerequisite: Agently >= 4.0.8.5.

Scope

Use this skill for:

• choosing between runtime_data / data.state and flow_data / data.flow_state
• deciding whether a dependency should be a flow-level or execution-level runtime resource
• data.set_resource(...) and execution-scoped resource injection
• understanding what execution save/load captures and what it does not
• understanding what flow-config export/import keeps and what it does not
• restart-safe state and resource design
• deciding how state and resources should cross a sub-flow boundary

Do not use this skill for: