I’m currently building a task orchestration system on top of Temporal, which will provide a unified operational interface (including pause, cancel for Activities) to upper-layer business systems (console, UI). I strongly agree with Temporal’s “Workflow as Code” philosophy.
However, the event-driven nature poses challenges for observability. While the Temporal Web UI doesn’t inherently show relationships between Activities (which aligns with its design), in reality, once a Workflow starts executing, its internal Activities do have dependencies, sequential relationships, parallel branches, etc. This execution graph (e.g., a DAG) is crucial for end-users to understand and monitor the system state.
Therefore, my question is: Are there any existing methods, tools, or open-source projects that can dynamically reconstruct the Activity execution dependency graph (DAG) by parsing a Workflow’s History Events? This would significantly enhance the observability of our system for the end-users.