Reuse and composition

Choose a reuse mechanism

Need	Mechanism
Run the same collection process again	Save the workflow and execute a new run
Continue from unchanged outputs	Use rerun controls and inspect what is reused
Apply a small workflow to each record	Inline forEach graph
Read earlier persisted results	Query, optionally scoped by execution and source nodes
Merge branches describing the same entity	Combine with an explicit shared key

Enrich a stored directory

A supported enrichment pattern

Query (records) → For Each (visit website → convert → extract)
                                      ↓
                                MongoDB Store → Query → Table

Give the Query a small limit. Bind each website into the inline fetcher’s url, set fetch depth to zero, and test a few items. Open foreach_result to check skipped and failed items before writing records at scale.

For multiple enrichment sources, connect record-producing branches directly to Combine, choose a stable key, and set field rules. Inspect unkeyed and _branches, then persist the merged records. See For Each, Combine, and storage.

Keep the abstraction honest

The implemented composition mechanism is an inline engine graph. Nested For Each nodes are rejected. A saved workflow is independently reusable by running it again, but this does not imply that another workflow can invoke it by id as a versioned function. No published reusable-subworkflow registry or SDK is asserted here.