What is Execution Engine?
Execution Engine records what your code did at runtime — inputs, outputs, errors and timing — without requiring you to rewrite function bodies. It runs in your process and returns plain data.
It has two independent modes: use Execution for one function, or Engine for a connected workflow.
Execution — individual functions
Capture calls independently, with no engine and no graph. Use Trace for call data, Cache or Memoize to avoid repeated work, and Timer for elapsed time.
Engine — a whole run
Route related calls through an engine. Each call becomes a node; edges capture order, nesting and parallel work. The result is a portable JSON Trace of the whole run.
Graph data that travels
Engine traces use a Cytoscape-compatible shape.
- Visualize — open a trace in the
json-to-graph viewer. - Test — assert directly on its nodes and edges.
- Resume — pass a stored trace back to an engine and continue the run.
Which one?
| Execution | Engine | |
|---|---|---|
| Scope | One function | Many calls, and their relationships |
| Output | One record per call, returned or sent to a callback | A JSON graph of nodes and edges |
| Setup | Nothing | An engine instance |
| Order, nesting, parallelism | Not tracked | Tracked |
| Use it for | Logging, metrics, debugging one function | Pipelines, workflows, orchestration |
The two are independent; use either without adopting the other.