Configure ManyRows workflows on your own schema
Can we enable workflows without installing a template?
Yes. Create your own types and fields, then open Project settings → Workflow behaviors and connect them to native roles. Templates package example schemas, explicit settings and optional fictional records. A familiar type or field name does not enable behavior.
Choose how to prepare your schema
Scroll sideways to compare
| Step | Independently authored schema | Example template |
|---|---|---|
| Create the model | Define your own record types, fields and relationships. | Install an example schema and adapt it to your work. |
| Connect behavior | Select a native profile and assign actual types and fields to its roles. | Review the explicit configuration packaged with the example. |
| Use real records | Create records through the configured workflow's own-data guide or ordinary record forms. | Fictional records illustrate the workflow; replace them with your own work. |
| Rename later | Saved connections follow immutable identities when keys or labels change. | The same identity rules apply after adapting the example. |
Connect the behaviors your team uses
Setup lists native record and field roles, compatible field kinds, reference targets and supported status selections. Required inputs appear first; optional connections are expandable and searchable. Unique compatible field suggestions still require your review. Current native profiles use the status values shown in setup.
- Open Project settings → Workflow behaviors and search for the behavior you need.
- Assign your active custom record types to the roles you use. Follow the purchasing supplier prerequisite where shown.
- Connect the required attached fields and any related target roles. Open Schema to add missing fields or selections, then refresh the choices.
- Review every connection and choose Save and enable. Use Save paused setup for incomplete work, then return through Finish setup.
Treat missing inputs as unfinished setup
An enabled setting alone does not establish readiness. REST and MCP can retain partial bindings; inspect the returned issues and requiredFields before using the affected behavior. Apply matching profile requirements, including their attachments on the named record roles, and expand requiresFields on connected field roles transitively too. A shared field on another record type cannot satisfy an owner-specific prerequisite. Their conditions use explicit role connections. These are native role identifiers, not customer keys. Missing or restricted evidence cannot prove that a work queue is clear.
Selecting an optional behavior reveals its prerequisites immediately. Report outcomes need physical observation inputs; finish selection needs its option list and finish specification. Unused optional behaviors add no requirements. Empty paused setup must be finished before it can resume. Supported native terminal selections such as Received or Approved can correctly leave a queue clear even when the field offers no active-work selections.
Furniture, electronics and robot model definition reviews connect their core inputs directly to the reviewed record type, including unit of measure and the relevant identifiers, description and lifecycle state. Setup shows these requirements without a separate parts or materials type or inherited example fields. Furniture definition inputs reveal their remaining review prerequisites when connected; lifecycle-only work queues stay valid. The Robotics development and service profile includes robot model reviews, commissioning and maintenance.
Recording forms need a type that permits direct creation and usable native defaults for required custom inputs they do not supply. Required references need native record entry rather than scalar defaults. Setup explains these constraints before enabling. Public type discovery exposes requiresCrOnCreate and effective field hasDefault flags; it does not expose default values.
For grocery private-label work, the configured artwork record type contributes its attached files to the personal review. File changes require another review, and unreadable files withhold derived progress. This works with independently named file fields; nested pack, recipe, specification and artwork links follow the saved type identities after renames.
Food sample observations need a subject reference, outcome, actual observation date, physical sample reference, method and findings. Connect all six roles even on a schema you wrote yourself. A missing outcome connection is unfinished setup, not evidence that no sample work remains.
When a schema edit makes a saved optional connection incompatible, workflow setup shows that connection and its repair message. Choose another compatible field, clear an unused optional connection, or update Schema and refresh. A deprecated type may need to be restored or replaced, subject to captured-history restrictions.
Enable related capabilities explicitly
Purchasing supplier eligibility is separate from allowing prices. Price inheritance selects explicit reference fields; without a selection, own prices apply. Sample review scope separately selects specimen subjects and physical size, with an optional derived-to-parent specification relationship. Operating hours has its own setup.
Commissioning requires release control on the selected declaration type. Setup links to Release control policy: enable change control on create and edit, then release control, and refresh the workflow choices. Catalog onboarding needs only a record type; pricing and channel capabilities remain separate.
Discover and review setup through REST or MCP
Read GET /types for type and attached field UUIDs, kinds, selections and reference targets. Read GET /schema/workflows for native role definitions, required inputs, saved versions and readiness issues. Bind discovered UUIDs through PUT /schema/workflows/{workflow}. GET /workflow-bindings resolves already saved identities; it does not establish readiness.
First workflow and sample-scope setup uses If-None-Match: *. Replacement uses the exact saved If-Match version. Price inheritance always uses its GET version, including an empty selection. After a 412 response, reread and explicitly review the current settings before saving again. Setup requires schema-management authority and complete field visibility; read-only managers can inspect it.
Start an observation workflow with your own names
For example, connect an Observation note type to the native food sample role and a Design note type to its subject role. Fields named Decision, Observation day, Physical unit, Procedure, Findings and Subject can supply the six required inputs when their kinds and targets match. No example field key or fictional record is required. Saved identities keep the connections through later renames.
Read the workflow setup API documentationWhere ManyRows fits
Configure one useful behavior first, review its readiness, and extend the schema as your work grows. Pausing keeps settings and captured history while blocking new governed operations. Configuration does not grant additional record, approval or certification permissions.
Common questions
Will naming a type supplier or sample enable a workflow?
No. Explicit capabilities and saved role bindings establish behavior. Customer names, schema keys and example field signatures do not.
Can we rename a connected field?
Yes. Connections follow native UUIDs. Changing a field's kind, attachment or reference target can require setup repair; renaming its key or label keeps its identity.
Does importing a template replace our settings?
Portable imports resolve source keys to destination identities, retain conflicting existing settings and report differences. Use reviewed setup operations when you intend to replace saved connections.