- Overview
- Get started
- Concepts
- Using UiPath CLI
- How-to guides
- CI/CD recipes
- Command reference
- Overview
- Exit codes
- Global options
- uip codedagent
- uip coder
- uip context-grounding
- uip docsai
- uip function
- uip guardrails
- uip llm-configuration
- uip llm-gateway
- uip maestro authoring
- uip maestro registry
- uip model-hub
- activities find
- activities get-default-xaml
- analyze
- analyzer-rules list
- build
- cleanup
- data-fabric-entities install
- data-fabric-entities list
- files diff
- get-library-object-repository
- get-object-repository
- init
- instances list
- pack
- packages inspect
- packages install
- packages versions
- publish
- remote
- restore
- run, debug & execution
- settings
- studio start
- templates search
- test-data add-entity
- test-data add-queue
- test-data add-variation
- tm
- uia
- validate
- workflow-examples get
- workflow-examples list
- uip tasks
- uip traces
- uip traces feedback
- Migration
- Reference & support
uip vss deploy seed
Syntax and options for `uip vss deploy seed`, which loads Data Fabric seed data for a deployed Vertical Solution from CSV files, and the CSV rules.
uip vss deploy seed loads a deployed Vertical Solution's Data Fabric seed data — reference data the package ships, such as rate tables, categories, or document templates. It reads one CSV file per entity from a seed workspace, validates the rows against the live entities, shows what changes, and writes the records. It runs after a successful uip vss deploy, either on its own or from the deploy with --seed.
Synopsis
uip vss deploy seed [options]
uip vss deploy seed [options]
The seed workspace
When the package declares seed entities, a successful watched deploy writes a seed workspace:
./seeding/<package-name>/
.seed-context.json where the deploy landed: package, version, and the
folder the solution installed into (not the target)
.seed-ids.json retry journal — do not delete
<Entity>.csv one per seeding entity; fill these in
<Entity>/<file> attachment files referenced by the CSVs
./seeding/<package-name>/
.seed-context.json where the deploy landed: package, version, and the
folder the solution installed into (not the target)
.seed-ids.json retry journal — do not delete
<Entity>.csv one per seeding entity; fill these in
<Entity>/<file> attachment files referenced by the CSVs
Right after a successful watched deploy, the CLI shows what the workspace holds. At a terminal, with neither --seed nor --no-seed, it asks Seed Data Fabric data now?. --seed answers that question for you, and --no-seed skips the offer.
--seed does not make the seed unattended: the seed still shows each entity's plan and asks before writing it, and a run with no terminal refuses and names --yes. For unattended seeding, pass --seed --yes.
Every watched deploy that succeeds writes the templates again, also with --no-seed, and replaces the CSV files already in the workspace. The run says (previous files replaced). Copy your filled-in CSV files to another location before you deploy again.
To seed later, fill in the CSV files, then run:
uip vss deploy seed
uip vss deploy seed
With no flags, the command uses the only workspace under ./seeding. When there is more than one, it asks which one at a terminal, and otherwise fails and asks for --package-name. The package, version, and folder then come from that workspace's .seed-context.json.
Options
| Option | Description |
|---|---|
--package-name <name> | Workspace to use (./seeding/<name>). Defaults to the only workspace under ./seeding. With more than one, the command asks at a terminal and requires the option otherwise. |
--package-version <version> | Version whose schema the CSV files are validated against. Defaults to the recorded version, which applies only to the package that recorded it. With --package-name, creates a missing workspace from that version's templates. |
--dir <path> | Folder that holds the CSV files. Defaults to ./seeding/<package-name>. |
--folder-path <path> | Folder the solution's Data Fabric entities live in, by path, for example Shared/Lending/LoanSetup. Overrides the recorded folder. Refused for a package whose entities are tenant-level (dataFabric.schemaFilename). |
--folder-key <guid> | The same folder, by key. Use either this option or --folder-path. |
-y, --yes | Seed every entity without asking. Required in a non-interactive terminal. |
--refresh | Download the templates again first. This replaces the CSV files and keeps the retry journal. .seed-context.json is kept too, unless the refresh is for another version or folder, in which case it records the new one. To seed a version other than the recorded one, pass --refresh. |
--login-validity <minutes> | Minutes of remaining token validity that trigger a token refresh. Default 10. |
Examples
# Seed Data Fabric using the workspace the last deploy wrote
uip vss deploy seed
# Seed a folder-scoped entity set from CI, without prompts
uip vss deploy seed --package-name <package-name> --package-version <version> \
--folder-key a1b2c3d4-e5f6-7890-abcd-ef1234567890 --yes
# Seed Data Fabric using the workspace the last deploy wrote
uip vss deploy seed
# Seed a folder-scoped entity set from CI, without prompts
uip vss deploy seed --package-name <package-name> --package-version <version> \
--folder-key a1b2c3d4-e5f6-7890-abcd-ef1234567890 --yes
What a seed run does
- Shows where the files are, which tenant and folder the data goes to, and how many rows each entity holds.
- Validates every CSV file against the live entities, and reports problems by entity, row, and column before it writes anything:
- Choice labels must exist in the choice set.
- Unique values and local
Idvalues must not repeat. - Attachment files must be present.
- Compares every row with Data Fabric and summarizes each entity as
+N insert, ~N update, N unchanged, then a total. An identical row writes nothing. A row that already exists but differs is handled according to the entity's seedingstrategyin the package:insert, the default, leaves it alone and only adds new rows;upsertupdates it, sending only the fields that differ. - If optional columns are empty in your file but hold values in Data Fabric, asks once whether to clear them. The default is no, and
--yesnever clears. This question covers relationship and attachment columns too, so a blank cell never drops a live relationship or deletes an uploaded file on its own. - At a terminal and without
--yes, asks per entity<Entity> has changes. Show diff?, and prints the diff only when you answer yes. The diff shows up to 8 inserts with their columns and up to 8 updates asfield: old → new, then… and N more. Then it asksSeed <Entity> (+N insert, ~N update)?. - Ends with
Seeding completed successfully.when it wrote a record, or with the list of problems and the command to retry.
Declining an entity in step 5 skips it, and also skips every entity whose rows reference it by local Id, down the chain. The run ends as a failure only when it holds back a dependent entity this way, and it names what it held back. Declining an entity that nothing depends on is not a failure. With --yes, every entity is approved and neither question is asked.
Some runs end differently:
- A retry that only uploads an attachment again reports that on the entity's own line and ends without the closing line, because no record changed.
No seeding CSV files with data found.— every template is still empty.All records are up to date. Nothing to seed.— nothing differs. When rows do differ but the entity's strategy isinsert, the line names them: existing records are left alone, which is why an edit did not land.
Which folder gets seeded
A solution installs into its own folder under the one you targeted, and a new-structure package's Data Fabric entities live there. The deploy records that folder in .seed-context.json, so a plain uip vss deploy seed finds it.
Older packages that declare dataFabric.schemaFilename keep their entities at tenant level, so their installed folder is not the seed scope. uip vss deploy refuses these packages, but if one was deployed with vertical-deploy-cli, you can still seed it with uip vss deploy seed.
Do not pass --folder-key: an explicit --folder-key wins over the manifest's tenant-level scope and puts the records into a folder. --folder-path is refused for these packages, with exit code 3.
uip vss deploy seed --package-name <package-name> --package-version 1.4.2 \
--yes --output json
uip vss deploy seed --package-name <package-name> --package-version 1.4.2 \
--yes --output json
To seed a different folder-scoped installation, pass --folder-path with the path Orchestrator shows, or --folder-key with its key — the InstalledFolderKey that uip vss deploy status <job-id> --output json reports for the job that installed it.
A workspace written by a deploy into a different organization or tenant is refused unless you name the folder yourself, so a copied folder cannot seed the wrong tenant.
If the run says the folder shows none of the entities to this login while Orchestrator lists them there, the login has no access to that folder. Ask for the folder role your vertical lists. For an external application, refer to Set up the external application.
Re-running is safe
.seed-ids.json records the Data Fabric record of every row an earlier run inserted, per organization, tenant, and folder. A seed that stopped halfway matches those rows on the next run instead of inserting them twice:
- Under the
upsertstrategy, it then updates them. - Under the default
insertstrategy, it leaves them alone. - A record whose attachment upload failed gets only the missing files.
Keep the file — --refresh preserves it.
Ctrl+C stops a seed before its next batch of rows or file upload, and the command exits 130. Every batch that landed is already in .seed-ids.json, with its files marked pending until they upload, so run the same command again to finish.
Under the insert strategy, the run again leaves the rows it already wrote alone, so a link to another row of the same entity that the stopped run had not set yet stays empty. When the seed ran from uip vss deploy --seed, the deploy keeps its own result, and the warning carries the command that finishes the seed.
How rows are matched
- A row with a value in its
Idcolumn is identified by that value, which is what.seed-ids.jsonrecords. - A row that leaves
Idempty is matched on the entity's unique fields or, when the entity has none, on every plain column it fills in. Editing such a row changes what it matches. - A relationship column can tell two otherwise identical records apart, but never decides on its own whether a record matches, so editing one does not make the row look new. The exception is a row whose only filled-in column is a relationship: with nothing else to match on, that link finds the row again.
Editing a column the match does not use is safe. Editing one it does use makes the row look new, and the default insert strategy then inserts it again. Give every row you expect to edit later an Id. Edits reach Data Fabric only for an entity whose strategy is upsert.
Attachments are compared by file name and size. When the login cannot read an attachment's name — Data Fabric returns only its ID to an external application — a file that is already attached counts as present and is not uploaded again. To remove it, leave the cell empty and answer yes when the seed asks whether to clear empty optional columns. --yes never clears.
Fill in the CSV files
| Cell | Meaning |
|---|---|
| Empty | Field omitted. Data Fabric applies its default. |
NULL | Explicit null. |
\NULL | The literal string NULL. |
| A choice label | Mapped to the choice-set value. Multi-value fields take comma-separated labels. |
A record ID, or the Id of a row in the referenced entity's CSV file | Relationship column. Referenced entities are seeded first, so local Id values resolve to the records they became. |
| A file name | Attachment column. Put the file in <Entity>/ inside the seed workspace — ./seeding/<package-name>/<Entity>/ by default, or <dir>/<Entity>/ when the run passes --dir <dir>. The cell holds the file name only. |
%{UIPATH_TENANT_ID}%, %{Solution.FolderPath}%, %{DataFabric.Entities['Other'].Id}% | Resolved before writing, from the login and from the folder this run seeds. Refer to Expressions. |
%{Solution.Processes['Name'].Id}%, %{Solution.Assets['Name'].Value}%, %{Solution.StorageBuckets['Name'].Id}%, %{Solution.Connections['Name'].Id}% | Read from the folder the solution installed into, with the login's own access, so an external application resolves them too. Refer to Expressions. |
Expressions
The login and folder expressions fall into four groups.
Platform, organization, and tenant:
UIPATH_BASE_URL(the platform URL)UIPATH_ORG_IDUIPATH_ORG_NAMEUIPATH_TENANT_IDUIPATH_TENANT_NAME
Folder, as UIPATH_FOLDER_*:
UIPATH_FOLDER_KEYUIPATH_FOLDER_IDUIPATH_FOLDER_PATH
Folder, as Solution.*:
Solution.FolderPathSolution.FolderKeySolution.FolderIdSolution.ParentFolderIdSolution.ParentFolderKey
Data Fabric entities:
DataFabric.Entities['Other'].IdDataFabric.Entities['Other'].Name
Solution.Processes, Solution.Assets, and Solution.Connections count only names that the manifest declares. Solution.StorageBuckets takes any bucket in the folder, or in a Solution folder directly under it. A name that is not found is written empty, with a warning. An expression that cannot be resolved is written empty, with one warning per entity that names the unresolved expressions.
Numbers and dates
- Numbers take plain decimal or exponent form (
12.5,1.25e1). Integer columns take plain digits only; hex, binary, and octal forms are refused. - The CLI sends a value that a double cannot hold exactly as the text you wrote: a value with more than 15 significant digits, a nonzero value too small to keep, or a
BIGINTpast 2^53. - Dates accept
YYYY-MM-DD,DD/MM/YYYY, andDD.MM.YYYY. Date/times also take aT- or space-separated time, with at most six digits of fractional seconds. - A zoned column (
DATETIMEOFFSET) also takes aZor+HH:MMoffset, and reads a bare timestamp as UTC. A wall-clock column (DATETIME,DATETIME2) holds no zone, so none is added, and a value that carries an offset is refused. The date and time are reformatted the same way in both kinds of column.
Attachments and relationships
- An attachment that already matches the live file, by name and size, is not uploaded again.
- An empty attachment cell on an existing record removes the live file only if you answer yes to the clear question in step 4 of What a seed run does.
--yesnever clears. - An optional relationship whose local
Idmatches no row and no record ID is written empty, with a warning that names the row. - A required relationship that matches nothing holds back the row that carries it, with an error, since the entity does not allow a record without it.
- A relationship to a row this run meant to write but could not is held back the same way, required or not.
Skipped entities
An entity whose CSV file is missing from the workspace is skipped with a warning and reported as Skipped in the output. Every skipped entity names why, in SkipReason:
SkipReason | Meaning |
|---|---|
no-seed-file | No CSV file in the workspace. |
not-in-scope | The scope has no such entity. |
invalid | Its rows did not plan. |
depends-on-skipped | It references an entity that was not seeded. |
declined | You answered no. |
A workspace that holds none of the declared files is refused, since it was never written. Run uip vss deploy seed --refresh to write the templates, or pass --dir if the files live elsewhere.
Data shape (--output json)
{
"Code": "VerticalSolutionSeeded",
"Data": {
"WorkspaceDir": "/Users/alice/seeding/<package-name>",
"Entities": [
{
"EntityName": "Loan",
"Inserted": 12,
"Updated": 0,
"Unchanged": 0,
"Ignored": 0
}
]
}
}
{
"Code": "VerticalSolutionSeeded",
"Data": {
"WorkspaceDir": "/Users/alice/seeding/<package-name>",
"Entities": [
{
"EntityName": "Loan",
"Inserted": 12,
"Updated": 0,
"Unchanged": 0,
"Ignored": 0
}
]
}
}
Exit codes
| Code | Meaning |
|---|---|
0 | The seed finished, including runs that had nothing to write. |
1 | An entity could not be written, or the run held back a dependent entity. |
2 | No usable login. |
3 | The input is not usable and no write was tried: a flag value such as --folder-path for a package with tenant-level entities, a CSV file that cannot be read, a workspace with none of the declared files, a federated entity, or entities that reference each other in a cycle. |
130 | Ctrl+C stopped the seed. Run the same command again to finish. |
A fault on the seed workspace folder is classified too. The CLI finds, locks, and creates that folder before it seeds. If this login cannot read or write it, the run fails with ErrorCode: local_permission_denied and Retry: RetryWillNotFix. Other faults at that step carry network_error and RetryLater.
For the exit codes shared by every command, refer to Exit codes.
Related
- uip vss deploy — deploy the package that writes the seed workspace.
uip df— inspect the Data Fabric entities and records the seed writes.