fix: Added advanced field option and offset support

skills/uipath-data-fabric/SKILL.md

@@ -130,6 +130,10 @@ uip df records query <entity-id> \
 
 Pass the exact `EntityFieldDataType` string — the CLI is case-sensitive. Common types: `STRING`, `INTEGER`, `DECIMAL`, `BOOLEAN`, `DATE`, `DATETIME`, `UUID`. For the full type table with SQL backing types, see [`references/entity-schema.md`](references/entity-schema.md).
 
+### Advanced Field Constraints
+
+Optional per-type constraints on create/update — `lengthLimit` (STRING, MULTILINE_TEXT), `maxValue` / `minValue` (INTEGER, BIG_INTEGER, DECIMAL, FLOAT, DOUBLE), `decimalPrecision` (DECIMAL, FLOAT, DOUBLE). See `references/entity-schema.md` for ranges and examples.
+
 ---
 
 ## Workflow: Discover → Act → Verify
@@ -151,7 +155,7 @@ uip df records list <entity-id> --cursor <NextCursor> --output json
 
 ## Query Request Format
 
-Pass via `--body` or `--file`. Use `--limit` and `--cursor` CLI flags for pagination — not body keys.
+Pass via `--body` or `--file`. Use `--limit`, `--cursor`, and `--offset` CLI flags for pagination — not body keys. See [Pagination](#pagination) below.
 
 ```json
 {
@@ -172,6 +176,10 @@ Pass via `--body` or `--file`. Use `--limit` and `--cursor` CLI flags for pagina
 - For `in` / `not in` use `"valueList": ["a","b","c"]` — **not** a comma-separated `value` string
 - Response includes `HasNextPage` and `NextCursor` — pass `NextCursor` to `--cursor` for the next page
 
+## Pagination
+
+`records list` / `records query` paginate via `--limit`, `--cursor`, `--offset`. See [`references/records-query.md`](references/records-query.md).
+
 ---
 
 ## Troubleshooting

skills/uipath-data-fabric/references/entity-schema.md

@@ -69,7 +69,8 @@ Both entity names and field names must:
   "isUnique": false,
   "isRbacEnabled": false,
   "isEncrypted": false,
-  "defaultValue": ""
+  "defaultValue": "",
+  "lengthLimit": 200
 }
 ```
 
@@ -84,6 +85,39 @@ Both entity names and field names must:
 | `isRbacEnabled` | boolean | `false` | Role-based access control on this field |
 | `isEncrypted` | boolean | `false` | Encrypted at rest |
 | `defaultValue` | string | — | Default value (always a string representation) |
+| `lengthLimit`, `maxValue`, `minValue`, `decimalPrecision` | number | type-specific | Advanced per-type constraints — see below |
+
+### Advanced Field Constraints
+
+Accepted on `entities create` and on `addFields` / `updateFields` in `entities update`. Each constraint applies only to specific types — passing one to an unsupported type errors with *"Field '<name>' of type <TYPE> does not accept <option>"*. `minValue` must be strictly less than `maxValue`.
+
+| Constraint | Allowed types | Range |
+|------------|---------------|-------|
+| `lengthLimit` | `STRING` (1–4000), `MULTILINE_TEXT` (1–10000) | — |
+| `maxValue` / `minValue` | `INTEGER`, `BIG_INTEGER`, `DECIMAL`, `FLOAT`, `DOUBLE` | ±9,007,199,254,740,991 |
+| `decimalPrecision` | `DECIMAL`, `FLOAT`, `DOUBLE` | 0–10 |
+
+```bash
+uip df entities create "Orders" \
+  --body '{
+    "fields": [
+      {"fieldName": "ProductName", "type": "STRING",  "lengthLimit": 500, "isRequired": true},
+      {"fieldName": "Price",       "type": "DECIMAL", "decimalPrecision": 4, "maxValue": 999999, "minValue": 0},
+      {"fieldName": "Quantity",    "type": "INTEGER", "maxValue": 10000, "minValue": 1}
+    ]
+  }' \
+  --output json
+```
+
+Change a constraint after creation via `updateFields` (use the field UUID from `entities get`):
+
+```bash
+uip df entities update <entity-id> \
+  --body '{"updateFields":[{"id":"<field-id>","lengthLimit":1000}]}' \
+  --output json
+```
+
+`entities get` echoes the current constraint values on each field under `Fields[].fieldDataType.{lengthLimit,maxValue,minValue,decimalPrecision}` — read these before authoring an `updateFields` call.
 
 ### Choice Set Fields
 
@@ -154,7 +188,7 @@ uip df entities update <entity-id> \
   --output json
 ```
 
-`updateFields` entry supports: `id` (required), `displayName`, `description`, `isRequired`, `isUnique`, `isRbacEnabled`, `isEncrypted`, `defaultValue`.
+`updateFields` entry supports: `id` (required), `displayName`, `description`, `isRequired`, `isUnique`, `isRbacEnabled`, `isEncrypted`, `defaultValue`, `lengthLimit`, `maxValue`, `minValue`, `decimalPrecision`. The four constraint keys follow the per-type allow-list in [Advanced Field Constraints](#advanced-field-constraints).
 
 ### Supported `entities update` Body Keys
 

skills/uipath-data-fabric/references/records-query.md

@@ -15,6 +15,24 @@ Response: `{ TotalCount, Records, HasNextPage, NextCursor?, CurrentPage?, TotalP
 - Use `HasNextPage` to check if more records exist
 - Pass the `NextCursor` string value to `--cursor` to fetch the next page
 
+## Pagination
+
+Offset-based under the hood. Available on both `records list` and `records query`:
+
+- `-l, --limit <number>` — page size, default `50`, min `1`. Keep constant across a sweep (changing it re-slices the offset and can skip/duplicate records).
+- `--cursor <NextCursor>` — opaque string from previous response. Pass verbatim; never hand-craft.
+- `-o, --offset <number>` — non-negative record index. Rounded down to the nearest page boundary (`jumpToPage = floor(offset / limit) + 1`). **Mutually exclusive with `--cursor`** — passing both errors with *"--offset and --cursor are mutually exclusive"*.
+- Stop when `HasNextPage: false`. `CurrentPage` / `TotalPages` are informational.
+
+```bash
+# Sequential sweep
+uip df records list <entity-id> --limit 100 --output json
+uip df records list <entity-id> --limit 100 --cursor "<NextCursor>" --output json
+
+# Jump directly to the page containing record #250 (with --limit 100 → page 3)
+uip df records list <entity-id> --limit 100 --offset 250 --output json
+```
+
 ## Filtered Query
 
 ```bash
@@ -23,7 +41,7 @@ uip df records query <entity-id> \
   --output json
 ```
 
-Pagination for query also uses `--limit` and `--cursor` flags — not body keys.
+Pagination for query also uses `--limit`, `--cursor`, and `--offset` flags — not body keys.
 
 ```bash
 # Query with pagination
@@ -52,7 +70,7 @@ uip df records query <entity-id> \
 }
 ```
 
-> `start` and `limit` are **not** valid body keys — use `--limit` and `--cursor` CLI flags instead.
+> Pagination is CLI-flag only: `--limit`, `--cursor`, `--offset`. Do not put `start`, `limit`, `offset`, or `pageSize` inside `--body` — the CLI translates these flags into the underlying offset-based API call.
 
 ### Operators
 

tests/tasks/uipath-data-fabric/integration_full_lifecycle.yaml

@@ -1,11 +1,15 @@
 task_id: skill-datafabric-integration-full-lifecycle
 description: >
   Integration test: agent performs a complete Data Fabric lifecycle against a
-  live tenant — verifies login, discovers or creates an entity, retrieves the
-  schema to collect field IDs from Fields[].ID, adds and updates fields, inserts
-  records, queries with a filter, updates one record, verifies the update with
-  records get, and deletes only the test records. Tests the full Discover → Act →
-  Verify workflow combining schema management and record CRUD in one multi-step run.
+  live tenant — verifies login, discovers or creates an entity with advanced
+  per-type field constraints (lengthLimit on STRING, minValue / maxValue on
+  INTEGER), retrieves the schema to collect field IDs and confirms the
+  constraints come back under Fields[].fieldDataType, adds and updates fields
+  (including tightening a constraint via updateFields), inserts records,
+  queries with a filter, updates one record, verifies the update with
+  records get, and deletes only the test records. Tests the full Discover →
+  Act → Verify workflow combining schema management, advanced field
+  constraints, and record CRUD in one multi-step run.
 tags: [uipath-data-fabric, integration, entities, records]
 
 agent:
@@ -25,19 +29,23 @@ initial_prompt: |
      whatever tenant is currently active.
 
   2. List native-only entities. If an entity named "LifecycleTest" already exists,
-     use its ID. If not, create it with fields:
-       - Label (STRING)
-       - Score (INTEGER)
+     use its ID. If not, create it with fields and advanced constraints:
+       - Label  (STRING,  lengthLimit 200)
+       - Score  (INTEGER, minValue 0, maxValue 1000)
        - Active (BOOLEAN)
 
   3. Retrieve the full schema for LifecycleTest using entities get.
      From the Fields array, note the ID (Fields[].ID) for the Score field.
+     Confirm the response echoes the constraints back under
+     Fields[].fieldDataType.{lengthLimit,minValue,maxValue} — print them so
+     they are visible in the transcript.
 
-  4. Add a new field to the entity: Notes (STRING).
+  4. Add a new field to the entity: Notes (MULTILINE_TEXT, lengthLimit 2000).
 
-  5. Update the Score field: set displayName to "Test Score" and isRequired to true.
-     Use the field ID from step 3 in the updateFields body — the payload key is
-     "id" (lowercase), not "ID".
+  5. Update the Score field: set displayName to "Test Score", isRequired to
+     true, and tighten maxValue from 1000 to 500. Use the field ID from
+     step 3 in the updateFields body — the payload key is "id" (lowercase),
+     not "ID".
 
   6. Insert two records:
        - { "Label": "alpha", "Score": 10, "Active": true }
@@ -169,6 +177,30 @@ success_criteria:
     weight: 2.0
     pass_threshold: 1.0
 
+  - type: command_executed
+    description: "Agent applied lengthLimit on a field (via entities create or addFields on update)"
+    tool_name: "Bash"
+    command_pattern: '(?s)uip\s+df\s+entities\s+(create|update).*lengthLimit'
+    min_count: 1
+    weight: 2.0
+    pass_threshold: 1.0
+
+  - type: command_executed
+    description: "Agent applied minValue/maxValue on a numeric field (via entities create or update)"
+    tool_name: "Bash"
+    command_pattern: '(?s)uip\s+df\s+entities\s+(create|update).*(minValue|maxValue)'
+    min_count: 1
+    weight: 2.0
+    pass_threshold: 1.0
+
+  - type: command_executed
+    description: "Agent tightened a numeric constraint via updateFields"
+    tool_name: "Bash"
+    command_pattern: '(?s)uip\s+df\s+entities\s+update.*updateFields.*(minValue|maxValue)'
+    min_count: 1
+    weight: 2.0
+    pass_threshold: 1.0
+
 post_run:
   - command: "python3 $SKILLS_REPO_PATH/tests/tasks/uipath-data-fabric/_shared/cleanup_entities.py"
     timeout: 60

tests/tasks/uipath-data-fabric/integration_large_pagination.yaml

@@ -2,10 +2,12 @@ task_id: skill-datafabric-integration-large-pagination
 description: >
   Integration test: agent imports 500 records from a CSV into a live entity,
   traverses all pages using cursor-based pagination (limit=60), verifies no
-  gap or duplicate record IDs across all 9 pages, then applies a filterGroup
-  query with pagination (limit=60) to verify the 300-record Active=true subset
-  across 5 pages. Covers large-dataset pagination stability and cursor
-  continuity at scale.
+  gap or duplicate record IDs across all 9 pages, jumps directly to a specific
+  page using -o/--offset and confirms the offset-page subset matches the
+  cursor-walked equivalent, then applies a filterGroup query with pagination
+  (limit=60) to verify the 300-record Active=true subset across 5 pages.
+  Covers large-dataset pagination stability, cursor continuity at scale, and
+  offset-based page jumps.
 tags: [uipath-data-fabric, integration, bulk-pagination]
 
 agent:
@@ -39,19 +41,46 @@ initial_prompt: |
   4. Import the CSV:
        uip df records import <entity-id> --file bulk_500.csv --output json
 
-  5. Traverse ALL records using cursor-based pagination (limit=60):
+  5. MANDATORY — Traverse ALL records using cursor-based pagination (limit=60).
+     This step MUST be completed in full before any later step. Do NOT skip,
+     summarise, or claim completion without running every page individually.
        - First call: uip df records list <entity-id> --limit 60 --output json
-       - Each response includes a cursor; pass it to the next call with --cursor
-       - Stop when no cursor is returned
-     Use Python to collect all record IDs across pages.
-     Expected: 9 pages (8 pages × 60 records + 1 page × 20 records = 500 total).
+       - For each subsequent call, pass the prior response's NextCursor verbatim
+         via --cursor, and continue to include --limit 60 --output json on
+         every call. Stop only when HasNextPage is false.
+       - Run each page as its OWN Bash invocation of `uip df records list`
+         (not via Python subprocess). Each page must show up as a separate
+         tool call.
+     Collect every record ID into a Python list named `cursor_walk_ids`.
+     Expected: 9 pages (8 pages × 60 records + 1 page × 20 records = 500
+     total) → 9 separate `uip df records list ... --limit 60` Bash calls.
 
   6. Use Python to verify:
        - total unique record IDs collected == 500
        - no duplicate IDs across pages
-
-  7. Query all Active=true records using records query with filterGroup and
-     paginate with limit=60 using --cursor until exhausted:
+     Print `len(cursor_walk_ids)` so the value is visible in the transcript.
+     If step 5 did not actually produce 9 separate tool calls, return to step 5
+     and finish the walk before continuing.
+
+  7. Only after step 6 has confirmed 500 IDs in `cursor_walk_ids`, jump to a
+     mid-range page using offset-based pagination:
+       uip df records list <entity-id> --limit 60 --offset 180 --output json
+     (--offset rounds down to the nearest page boundary, so offset 180 with
+     limit 60 lands on page 4, returning records 181..240 of the entity.)
+     Save the returned record IDs as `offset_page_ids`.
+
+     Use Python to confirm:
+       - exactly 60 records came back
+       - those 60 IDs match records 181..240 of the full cursor-walked list
+         (i.e. offset_page_ids == cursor_walk_ids[180:240])
+
+     Do NOT pass --offset and --cursor on the same command — they are
+     mutually exclusive.
+
+  8. MANDATORY — Query all Active=true records using records query with
+     filterGroup and paginate with limit=60 using --cursor until exhausted.
+     This step MUST run; each page is its own `uip df records query` Bash
+     call:
        {
          "filterGroup": {
            "logicalOperator": 0,
@@ -61,7 +90,7 @@ initial_prompt: |
      Expected: 5 pages × 60 records = 300 records total.
      Collect all matching record IDs.
 
-  8. Use Python to verify:
+  9. Use Python to verify:
        - every collected record has Active=true
        - total count == 300
 
@@ -92,6 +121,23 @@ success_criteria:
     weight: 3.0
     pass_threshold: 1.0
 
+  - type: command_executed
+    description: "Agent jumped to a page using -o/--offset on records list"
+    tool_name: "Bash"
+    command_pattern: 'uip\s+df\s+records\s+list\s+.*(-o\s+\d+|--offset\s+\d+)'
+    min_count: 1
+    weight: 2.5
+    pass_threshold: 1.0
+
+  - type: command_executed
+    description: "Agent must NOT combine --offset and --cursor on the same command"
+    tool_name: "Bash"
+    command_pattern: 'uip\s+df\s+records\s+(list|query)(?=[^\n]*--cursor)(?=[^\n]*(-o\s+\d+|--offset\s+\d+))'
+    min_count: 0
+    max_count: 0
+    weight: 2.0
+    pass_threshold: 1.0
+
   - type: command_executed
     description: "Agent queried Active=true records with filterGroup"
     tool_name: "Bash"