mammoth-cli 2.0.28__tar.gz → 2.0.29__tar.gz

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mammoth-cli
-Version: 2.0.28
+Version: 2.0.29
 Summary: Command-line interface for the Mammoth Analytics platform
 Project-URL: homepage, https://mammoth.io
 Project-URL: repository, https://github.com/EdgeMetric/mammothsdk

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/__init__.py

@@ -6,7 +6,7 @@ not implement a second HTTP client and does not call private SDK members.
 
 from __future__ import annotations
 
-__version__ = "2.0.28"
+__version__ = "2.0.29"
 
 # The versioned machine output envelope contract.
 SCHEMA_VERSION = 1

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/bundled_skill/mammoth-cli/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: mammoth-cli
-version: 2.0.28
+version: 2.0.29
 description: "Use Mammoth Analytics from a terminal: install or authenticate the CLI, discover its live command contract, and safely manage projects, data, views, pipelines, dashboards, exports, and handoffs."
 ---
 

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/bundled_skill/mammoth-cli/references/capabilities-misc.md

@@ -140,4 +140,4 @@ Ran once: `workflow.create`, `workflow.delete`, `workflow.get`, `workflow.graph`
 
 Ran once: `workspace.app-usage`, `workspace.get`, `workspace.list`, `workspace.segment.list`, `workspace.segment.update`, `workspace.storage-breakdown`, `workspace.user.list`
 
-Evidence collected on CLI releases 1.1.5 through 2.0.27; each row's release is recorded in `docs/release-capability-matrix.json` (`evidence_version`). A row that ran on an older release has not been re-run since unless its note says so. Details: `docs/capability-evidence/` in the repository.
+Evidence collected on CLI releases 1.1.5 through 2.0.28; each row's release is recorded in `docs/release-capability-matrix.json` (`evidence_version`). A row that ran on an older release has not been re-run since unless its note says so. Details: `docs/capability-evidence/` in the repository.

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/bundled_skill/mammoth-cli/references/capabilities.md

@@ -118,4 +118,4 @@ Ran once: `view.active-user.list`, `view.active-user.mark`, `view.ai.generate-da
 | `view.data-check.update` | observed blocker | backend_error: PATCH with the schema's documented {op: command, path: enable/disable, value: null} shape returns HTTP 500 empty body (CLI outcome_unknown, exit 7) while the mutatio |
 | `view.export.publish-db-update` | CLI defect fixed, untried since | 2.0.16 (example value is the documented {"odbc_type": "postgres"} object (a bare string is rejected)) |
 
-Evidence collected on CLI releases 1.1.5 through 2.0.27; each row's release is recorded in `docs/release-capability-matrix.json` (`evidence_version`). A row that ran on an older release has not been re-run since unless its note says so. Details: `docs/capability-evidence/` in the repository.
+Evidence collected on CLI releases 1.1.5 through 2.0.28; each row's release is recorded in `docs/release-capability-matrix.json` (`evidence_version`). A row that ran on an older release has not been re-run since unless its note says so. Details: `docs/capability-evidence/` in the repository.

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/bundled_skill/mammoth-cli/references/commands/dashboard.md

@@ -70,7 +70,7 @@ Example: `mammoth dashboard canvas get 123`. Placeholders are illustrative; reso
 
 Result: `DashboardCanvasGetResult`; mutation `read`, confirmation `none`, wait policy `not_async`.
 
-Status on release: ran once on CLI 2.0.12 — Published PyPI CLI 1.1.11 / SDK 0.7.1 draft canvas read succeeded for retained dashboard 48. One dashboard only; not Full. Also: Dashboard sweep 2026-09-18: exit 0 on release with CLI 2.0.12; result keys: canvas, dashboard_id, meta, plan, specs. returned canv…
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. Read back the authored canvas, plan.hints.kpis and meta.figures (figure key -> descriptor id) after each bake; used as the binding check the ILG brief requires. Single invocation only.
 
 ### `dashboard.canvas.restore`
 
@@ -90,7 +90,7 @@ Example: `mammoth dashboard canvas save 123 --input '{"body": {"params": {"canva
 
 Result: `DashboardCanvasSaveResult`; mutation `benign_mutation`, confirmation `none`, wait policy `not_async`.
 
-Status on release: ran once on CLI 2.0.14 — dashboard re-verification 2026-09-18: exit 0 on release with CLI 2.0.14. CLI defect fixed, confirmed: dashboard canvas get -> save round-trip completed cleanly, sequence advanced 1->2, response {"bake_job_id":296,"changed_tiles":0,"recomposed":false,"seeded_p…
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. Authored a canvas from a blank one: page-level focus.kpis (sum with unit.prefix GBP, and a countDistinct card), added bar (measure+measure2), hbar, line on a derived ratio measure (canvas.derived),…
 
 ### `dashboard.chat.edit`
 
@@ -178,7 +178,7 @@ Example: `mammoth dashboard create-blank --input '{"params": {"dataview_id": 1}}
 
 Result: `DashboardCreateBlankResult`; mutation `benign_mutation`, confirmation `none`, wait policy `not_async`.
 
-Status on release: ran once on CLI 2.0.21 — ergonomics sweep 2026-09-19: exit 0 on release with CLI 2.0.21. Created dashboard 62 (sequence 1). Single invocation only.
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. Two blank v3 dashboards bound to owned views (ids 63, 64). Single invocation only.
 
 ### `dashboard.data.draft`
 
@@ -214,11 +214,11 @@ Status on release: ran once on CLI 2.0.21 — ergonomics sweep 2026-09-19: exit
 
 Run: `mammoth dashboard descriptor-data`. Exact input fields: `mammoth schema get dashboard.descriptor-data`.
 
-Example: `mammoth dashboard descriptor-data 123 --input '{"body": {"params": {"descriptor_ids": ["resource-123"]}}}'`. Placeholders are illustrative; resolve IDs and input from observed reads.
+Example: `mammoth dashboard descriptor-data 123 --input '{"body": {"params": {"descriptor_ids": ["ef28f9bcc263c0f7"], "filter_state": {"Month": ["2025-01-01", "2025-03-31"]}}}}'`. Placeholders are illustrative; resolve IDs and input from observed reads.
 
 Result: `DashboardDescriptorDataResult`; mutation `benign_mutation`, confirmation `none`, wait policy `always_wait`.
 
-Status on release: ran once on CLI 2.0.12 — Dashboard sweep 2026-09-18: exit 0 on release with CLI 2.0.12; result keys: completed, expected, results. endpoint works on v3 engine; returned structured per-id error (404 unknown descriptor id) for a fabricated descriptor id since no real descriptor id was…
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. Evaluated baked descriptors by id with filter_state {Month: [start, end]}: countDistinct card = 111 unfiltered and 64 under a Q1 range (the fixture key; a summed per-cell count would read 83), sums…
 
 ### `dashboard.duplicate`
 
@@ -258,7 +258,7 @@ Example: `mammoth dashboard get 123`. Placeholders are illustrative; resolve IDs
 
 Result: `DashboardGetResult`; mutation `read`, confirmation `none`, wait policy `not_async`.
 
-Status on release: ran once on CLI 2.0.12 — Published PyPI CLI 1.1.11 read-only dashboard get on retained dashboard48 returned a bounded structural object; one retained scope does not establish Full support. Also: Dashboard sweep 2026-09-18: exit 0 on release with CLI 2.0.12; result keys: auto_publish,…
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. auto_sync for the bound view showed last_synced_at after the upstream data change. Single invocation only.
 
 ### `dashboard.get-by-url`
 
@@ -324,11 +324,11 @@ Status on release: ran once on CLI 2.0.12 — Dashboard sweep 2026-09-18: exit 0
 
 Run: `mammoth dashboard pages add`. Exact input fields: `mammoth schema get dashboard.pages.add`.
 
-Example: `mammoth dashboard pages add 123 --input '{"body": {"params": {"pages": [{}]}}}'`. Illustrative only: append `--yes --confirm <EXACT_TARGET>` after observing the target.
+Example: `mammoth dashboard pages add 123 --input '{"body": {"params": {"pages": [{"title": "By region", "focus": {"measure": "Revenue", "dim": "Region", "kpis": [{"field": "Revenue", "agg": "sum", "label": "Revenue"}]}, "charts": [{"kind": "hbar", "title": "Revenue by region", "dim": "Region", "measure": "Revenue", "agg": "sum"}]}]}}}'`. Illustrative only: append `--yes --confirm <EXACT_TARGET>` after observing the target.
 
 Result: `DashboardPagesAddResult`; mutation `high_impact`, confirmation `confirm_target`, wait policy `returns_job`.
 
-Status on release: ran once on CLI 2.0.12 — Dashboard sweep 2026-09-18: exit 0 on release with CLI 2.0.12; result keys: added_page_ids, bake_job_id, message, sequence. added page p2 to D=52 after supplying --yes --confirm 52 (first attempt without confirmation correctly returned confirmation_required,…
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. Needs --yes --confirm DASHBOARD_ID. body.params.pages[] takes PageAdd objects (title, focus{kpis}, charts[]); the route runs through the LLM guard and dropped unit.prefix on a money card ('nothing…
 
 ### `dashboard.pdf-artifact`
 

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/bundled_skill/mammoth-cli/references/commands/dataset.md

@@ -130,7 +130,7 @@ Example: `mammoth dataset list`. Placeholders are illustrative; resolve IDs and
 
 Result: `DatasetListResult`; mutation `read`, confirmation `none`, wait policy `not_async`.
 
-Status on release: ran once on CLI 2.0.18 — fixture-lifecycle sweep 2026-09-19: exit 0 on release with CLI 2.0.18. Empty result on this fixture. Confirmed all 3 owned datasets (85,86,87) gone before deleting the project. Single invocation only.
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. Listed the datasets the sends created in the target project. Single invocation only.
 
 ### `dataset.rename`
 

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/bundled_skill/mammoth-cli/references/commands/file.md

@@ -82,7 +82,7 @@ Example: `mammoth file upload ./sales.csv`. Placeholders are illustrative; resol
 
 Result: `FileUploadResult`; mutation `benign_mutation`, confirmation `none`, wait policy `start_or_wait`.
 
-Status on release: ran once on CLI 2.0.21 — ergonomics sweep 2026-09-19: exit 0 on release with CLI 2.0.21. Uploaded smoke.csv -> dataset 94 ready; dataview 116 (3 rows). Second call with --input {"append_to_ds_id": 94} returned dataset 95 ready and view 116 row_count went 3 -> 6: append path works (th…
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. Three CSV uploads into a new source project; dates typed DATE without file-settings changes. Single invocation only.
 
 ### `file.upload-folder`
 

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/bundled_skill/mammoth-cli/references/commands/view.md

@@ -220,7 +220,7 @@ Example: `mammoth view data get 123 123`. Placeholders are illustrative; resolve
 
 Result: `ViewDataGetResult`; mutation `read`, confirmation `none`, wait policy `always_wait`.
 
-Status on release: ran once on CLI 2.0.21 — ergonomics sweep 2026-09-19: exit 0 on release with CLI 2.0.21. Default trim: rows_returned 3, rows_total_in_page 3, truncated false; with limit 2: rows_returned 2, truncated true. Single invocation only.
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. Paged reads (limit up to 500) used for every value check; envelope data.data rows keyed by display name. Single invocation only.
 
 ### `view.data.query`
 
@@ -466,7 +466,7 @@ Example: `mammoth view export list 123 123`. Placeholders are illustrative; reso
 
 Result: `ViewExportListResult`; mutation `read`, confirmation `none`, wait policy `not_async`.
 
-Status on release: ran once on CLI an earlier release — Published PyPI CLI 1.1.11 read-only export-list call on retained view46/dataset29 returned a bounded empty page with unexpected offset50; no export was executed and Full support is not claimed.
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. Listed both internal_dataset exports with target_properties (DS_NAME, COLUMN_MAPPING, project fields). Single invocation only.
 
 ### `view.export.managed-s3`
 
@@ -702,7 +702,7 @@ Example: `mammoth view pipeline items 123`. Placeholders are illustrative; resol
 
 Result: `ViewPipelineItemsResult`; mutation `read`, confirmation `none`, wait policy `not_async`.
 
-Status on release: ran once on CLI 1.1.11 — Published PyPI CLI 1.1.11 pipeline-items read on retained view46/dataset29 returned one bounded item; prior owned-view evidence is retained and no Full claim is made.
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. fields=__full listed both export steps with sequence/status/execution times; the pre-existing step's timestamps did not move when a second send was appended. Single invocation only.
 
 ### `view.pipeline.items-all`
 
@@ -772,7 +772,7 @@ Example: `mammoth view task delete 123 123`. Illustrative only: append `--yes` a
 
 Result: `ViewTaskDeleteResult`; mutation `destructive`, confirmation `prompt_or_yes`, wait policy `always_wait`.
 
-Status on release: ran once on CLI 2.0.14 — write sweep 2026-09-18: exit 0 on release with CLI 2.0.14. Task 24 (created via view transform convert-type) deleted -> {future_id:347,status:processing,type_of_modification:discard_rule}. Verified via view task list 73: tasks:[] afterward. Single invocation…
+Status on release: ran once on CLI 2.0.28 — ILG simulation 2026-09-19: exit 0 on release with CLI 2.0.28. Removed a SQL task that had run on the wrong intermediate shape; view returned to the previous columns and row count. Single invocation only.
 
 ### `view.task.get`
 
@@ -832,7 +832,7 @@ Example: `mammoth view transform add-sql 123 --input '{"query": "SELECT region,
 
 Result: `ViewTransformAddSqlResult`; mutation `reversible_pipeline`, confirmation `none`, wait policy `always_wait`.
 
-Status on release: untried; it submits through `view.task.add`, but this transform was not among those run.
+Status on release: ran once on CLI 2.0.18 through `view.task.add` and was read back with `view data get`; other inputs for this transform are untried.
 
 ### `view.transform.ai`
 

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/bundled_skill/mammoth-cli/references/input.md

@@ -10,8 +10,8 @@ mammoth view transform filter 1039 --project 180 \
   --input '{"condition": {"and": [{"column": "Order Status", "operator": "EQ", "value": "Open"}, {"column": "Customer Age", "operator": "GT", "value": 30}]}}'
 ```
 
-- `--input FILE` reads a JSON or YAML file; the format is inferred from the
-  extension.
+- `--input FILE` (or `--input @FILE`) reads a JSON or YAML file; the format
+  is inferred from the extension.
 - `--input -` reads stdin; then `--input-format json|yaml` is required.
 - The top level must be a mapping. A bad path, format, or shape fails with exit
   code 2 and a stable error code.

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/bundled_skill/mammoth-cli/references/recipes/dashboards.md

@@ -34,6 +34,54 @@ Page/widget schemas vary by release. Read each schema, use returned IDs, and
 verify the binding plus draft/published data after every mutation. A
 creation response alone is not proof of a usable published view.
 
+## Authoring a board from a blank canvas
+
+`create-blank` gives one empty page. Author it by reading the canvas,
+editing the **active page** (`pages[0]`; a root-level `focus` is stored but
+never baked when pages exist) and writing the whole object back:
+
+```bash
+mammoth dashboard create-blank --yes --input '{"params": {"dataview_id": VIEW_ID, "title": "Overview"}}'
+mammoth dashboard canvas get DASHBOARD_ID > canvas.json      # data.canvas is the object to edit
+# pages[0].focus  = {"measure": "Sales Actual", "dim": "Location",
+#                    "kpis": [{"field": "Sales Actual", "agg": "sum", "label": "Sales", "unit": {"prefix": "£"}, "decimals": 0},
+#                             {"field": "Employee Ref", "agg": "countDistinct", "label": "Distinct staff"}]}
+# pages[0].added  = [{"kind": "bar", "title": "Actual vs Budget by Month", "measure": "Sales Actual", "measure2": "Sales Budget",
+#                     "agg": "sum", "date_bucket": {"field": "Month", "unit": "month"}},
+#                    {"kind": "hbar", "title": "Sales by Location", "dim": "Location", "measure": "Sales Actual", "agg": "sum", "sort": "desc"},
+#                    {"kind": "line", "title": "Occupancy rate", "measure": "occupancy_rate", "date_bucket": {"field": "Month", "unit": "month"}},
+#                    {"kind": "table", "title": "Detail", "columns": ["Location", "Month", "Places"], "sort_by": "Month", "limit": 100}]
+# derived         = [{"id": "occupancy_rate", "label": "Occupancy rate", "numerator": "Occupied", "denominator": "Places"}]
+# filters         = [{"field": "Month", "control": "range", "label": "Month"}, {"field": "Location", "control": "multi"}]
+mammoth dashboard canvas save DASHBOARD_ID --input '{"body": {"params": {"canvas": <edited data.canvas>}}}'
+mammoth job wait BAKE_JOB_ID                                  # data.bake_job_id from the save
+```
+
+Rules the backend enforces (each returns the pydantic path on failure):
+`added[].measure` is a column name — a ratio lives in `canvas.derived[]` and
+is referenced by id; `focus.kpis[].agg` accepts `countDistinct` (the only
+correct aggregation for a "distinct people" card under a date filter);
+`unit.prefix` goes on money cards only. `dashboard pages add --yes --confirm
+ID --input '{"body": {"params": {"pages": [{"title": ..., "focus": {...},
+"charts": [...]}]}}}'` adds a page, but that route runs through the LLM
+guard and may drop a unit it cannot evidence (it says so in `data.message`);
+`canvas save` keeps what you wrote.
+
+**Read the bindings back, then the numbers.** `dashboard canvas get` returns
+`data.meta.figures` — `"p1:kpi:2": {"descriptors": {"value": "<id>"}}` per
+card and tile — and `data.plan.hints.kpis` with the agg per card. Evaluate
+ids with a filter:
+
+```bash
+mammoth dashboard descriptor-data DASHBOARD_ID --input '{"body": {"params": {
+  "descriptor_ids": ["<id>", "<id>"],
+  "filter_state": {"Month": ["2025-01-01", "2025-03-31"]}}}}'
+# -> data.results.<id>.value (scalar) or .data (rows); filter fields must be declared canvas filters
+```
+
+A card that reads the same with and without the filter, or a distinct count
+that equals a row count, is a wrong binding, not a data fact.
+
 ## Canvas, widget data and PDF
 
 Never save an invented canvas. Read it, change it, write it back:

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/bundled_skill/mammoth-cli/references/recipes/transforms.md

@@ -142,6 +142,18 @@ Prefer `pivot` over `add-sql` for a grouped summary when `capabilities.md`
 lists it as run on release; `add-sql` replaces every column of the view, so
 never run it on a deliverable view.
 
+`add-sql` facts from release: the only table is `"view:VIEW_ID"` (the view
+itself; a second view in FROM/JOIN is `5GENR010 only one table allowed` —
+join with `view transform join` afterwards); `CASE`, `DATE_TRUNC('month',
+col)`, `GROUP BY`, `COUNT(DISTINCT col)` and `UNION ALL` over the same view
+work; result columns must be NUMERIC, TEXT or TIMESTAMPTZ (`CAST(x AS DATE)`
+is rejected; `CAST(x || '-01' AS TIMESTAMPTZ)` turns a `YYYY-MM` text month
+into a date column). A location canonicaliser plus month bucketing plus
+per-cell aggregation is one `add-sql` per fact; a mixed-grain table (summary
+rows plus one row per entity with NULL measures, for a `countDistinct` card)
+is one `UNION ALL`. The worked run is
+`docs/capability-evidence/ilg-sim-20260919/SUMMARY.md`.
+
 Reject malformed fields with the structured error envelope and stop rather than
 guessing names. Verify exact display names, join multiplicity, math values and
 remote task/pipeline readback. Draft submit is not proof all tasks persisted:

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/runtime/input_loader.py

@@ -157,6 +157,18 @@ def _read_bytes(input_file: str) -> bytes:
         ) from exc
 
 
+def _strip_at_prefix(input_file: str) -> str:
+    """Accept the ``@path`` file convention of curl and gh.
+
+    A value that starts with ``@`` and names an existing file is read from
+    that file; anything else (inline JSON, ``-``, a plain path) is returned
+    unchanged so an error still quotes what the caller typed.
+    """
+    if len(input_file) > 1 and input_file.startswith("@") and Path(input_file[1:]).is_file():
+        return input_file[1:]
+    return input_file
+
+
 def _read_text(input_file: str) -> str:
     """Read and strictly decode one input payload.
 
@@ -348,6 +360,7 @@ def load_input_document(input_file: str | None, input_format: str | None) -> dic
     """
     if input_file is None:
         return None
+    input_file = _strip_at_prefix(input_file)
     fmt = _resolve_format(input_file, input_format)
     document = _parse(_read_text(input_file), fmt)
     if not isinstance(document, dict):

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/mammoth_cli/services/input_fields.py

@@ -288,6 +288,43 @@ _EXAMPLE_INPUT_HINTS: dict[str, dict[str, Any]] = {
     },
     # The web app posts {condition, limit: null}; an empty body is 4GENR007.
     "view.derivative.data": {"body": {"limit": None}},
+    # pages[] takes PageAdd objects; the bare {} the schema suggests adds an
+    # empty page. focus.kpis and charts use the canvas vocabulary.
+    "dashboard.pages.add": {
+        "body": {
+            "params": {
+                "pages": [
+                    {
+                        "title": "By region",
+                        "focus": {
+                            "measure": "Revenue",
+                            "dim": "Region",
+                            "kpis": [{"field": "Revenue", "agg": "sum", "label": "Revenue"}],
+                        },
+                        "charts": [
+                            {
+                                "kind": "hbar",
+                                "title": "Revenue by region",
+                                "dim": "Region",
+                                "measure": "Revenue",
+                                "agg": "sum",
+                            }
+                        ],
+                    }
+                ]
+            }
+        }
+    },
+    # Ids come from dashboard canvas get -> data.meta.figures; filter_state
+    # keys are declared canvas filters (range control: [start, end]).
+    "dashboard.descriptor-data": {
+        "body": {
+            "params": {
+                "descriptor_ids": ["ef28f9bcc263c0f7"],
+                "filter_state": {"Month": ["2025-01-01", "2025-03-31"]},
+            }
+        }
+    },
     # The backend accepts NUMERIC, TEXT or DATE (4PARM008 otherwise).
     "parameter.create": {"param_type": "TEXT", "value": "Q3"},
     # The backend requires a placement ("Data check position must be given");

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "mammoth-cli"
-version = "2.0.28"
+version = "2.0.29"
 description = "Command-line interface for the Mammoth Analytics platform"
 authors = [{name = "Mammoth Analytics", email = "support@mammoth.io"}]
 readme = "README.md"

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/spec/manifests/commands/dashboard.yaml

@@ -358,7 +358,7 @@ commands:
   - CT-DASHBOARD-CANVAS-SAVE-ERROR
   draft_test: null
   undo_test: null
-  known_restrictions: null
+  known_restrictions: 'Read the canvas with dashboard canvas get, edit, write back; a canvas with pages[] bakes the ACTIVE page (pages[i].focus / pages[i].added), a root-level focus is stored but produces no descriptors. focus.kpis entries are {field, agg, label, unit, decimals, filters}; agg countDistinct persists and bakes (ILG simulation 2026-09-19). added[].measure is a column name; a ratio goes in canvas.derived[] as {id, numerator, denominator} and is referenced by id from measure or a KPI field. Validation errors return the pydantic path (e.g. pages.0.added.2.measure). The response carries bake_job_id: job wait it, then dashboard canvas get for meta.figures (figure key -> descriptor id) and dashboard descriptor-data to evaluate. This raw route does not pass the LLM unit guard that pages add / chat edit apply.'
   reviewed_by: primary
 - command_id: dashboard.chat.edit
   command_path: dashboard chat edit
@@ -701,7 +701,7 @@ commands:
   - CT-DASHBOARD-CREATE-BLANK-ERROR
   draft_test: null
   undo_test: null
-  known_restrictions: null
+  known_restrictions: 'Input is {"params": {"dataview_id": VIEW_ID, "title": ...}} with --yes; returns {id, sequence}. The blank canvas has one page (p1) and no figures: author it with dashboard canvas get -> edit -> dashboard canvas save or dashboard pages add. One dataview per dashboard.'
   reviewed_by: primary
 - command_id: dashboard.data.draft
   command_path: dashboard data draft
@@ -854,8 +854,7 @@ commands:
   required_fixture_guard: null
   secret_fields: []
   human_example: mammoth dashboard descriptor-data DASHBOARD_ID --help
-  agent_example: 'mammoth dashboard descriptor-data 123 --input ''{"body": {"params": {"descriptor_ids":
-    ["resource-123"]}}}'''
+  agent_example: 'mammoth dashboard descriptor-data 123 --input ''{"body": {"params": {"descriptor_ids": ["ef28f9bcc263c0f7"], "filter_state": {"Month": ["2025-01-01", "2025-03-31"]}}}}'''
   unit_tests:
   - UT-DASHBOARD-DESCRIPTOR-DATA
   contract_tests:
@@ -864,7 +863,7 @@ commands:
   - CT-DASHBOARD-DESCRIPTOR-DATA-ERROR
   draft_test: null
   undo_test: null
-  known_restrictions: null
+  known_restrictions: 'Evaluates baked descriptors by id. Ids come from dashboard canvas get data.meta.figures[<page>:<kpi|add>:<n>].descriptors.value (rows for a table figure) and data.meta.filters. body.params.filter_state is {"<filter field>": value} against the canvas''''s declared filters: a range control takes [start, end] (dates as YYYY-MM-DD), categorical controls a list of values; an undeclared field is rejected. Results are {descriptor_id: {status, value | data}}. Used to prove a countDistinct card under a Month range (ILG simulation 2026-09-19).'
   reviewed_by: primary
 - command_id: dashboard.duplicate
   command_path: dashboard duplicate
@@ -1290,7 +1289,7 @@ commands:
   required_fixture_guard: null
   secret_fields: []
   human_example: mammoth dashboard pages add DASHBOARD_ID --help
-  agent_example: 'mammoth dashboard pages add 123 --input ''{"body": {"params": {"pages": [{}]}}}'''
+  agent_example: 'mammoth dashboard pages add 123 --input ''{"body": {"params": {"pages": [{"title": "By region", "focus": {"measure": "Revenue", "dim": "Region", "kpis": [{"field": "Revenue", "agg": "sum", "label": "Revenue"}]}, "charts": [{"kind": "hbar", "title": "Revenue by region", "dim": "Region", "measure": "Revenue", "agg": "sum"}]}]}}}'''
   unit_tests:
   - UT-DASHBOARD-PAGES-ADD
   contract_tests:
@@ -1299,7 +1298,7 @@ commands:
   - CT-DASHBOARD-PAGES-ADD-ERROR
   draft_test: null
   undo_test: null
-  known_restrictions: null
+  known_restrictions: 'Requires --yes --confirm DASHBOARD_ID. body.params.pages is a list of PageAdd objects: {title, focus: {measure, dim, kpis: [...]}, charts: [{kind, title, dim, measure, agg}]}; the schema example''''s {} entry is not a usable page. The route runs through the LLM control guard: in the ILG simulation it dropped unit.prefix on a money card ("nothing in this data says which currency") that dashboard canvas save had kept, and reports that in data.message.'
   reviewed_by: primary
 - command_id: dashboard.pdf-artifact
   command_path: dashboard pdf-artifact

{mammoth_cli-2.0.28 → mammoth_cli-2.0.29}/spec/manifests/commands/view.yaml

@@ -3708,7 +3708,7 @@ commands:
   - CT-VIEW-TRANSFORM-ADD-SQL-ERROR
   draft_test: LT-VIEW-TRANSFORM-ADD-SQL-DRAFT
   undo_test: LT-VIEW-TRANSFORM-ADD-SQL-UNDO
-  known_restrictions: null
+  known_restrictions: 'Single-view SQL: the only table is "view:VIEW_ID" (the view the task is added to); a second view in FROM or JOIN is rejected (5GENR010 "only one table allowed"), use view transform join for that. UNION ALL over the same view, CASE, DATE_TRUNC, GROUP BY and COUNT(DISTINCT ...) run on release. Result columns must be NUMERIC, TEXT or TIMESTAMPTZ: CAST(x AS DATE) is rejected (5GENR010 "Invalid column type DATE"); CAST(x AS TIMESTAMPTZ) or DATE_TRUNC land as a DATE column. The task replaces every column of the view; run it on a copy (view export dataset) when the source columns are still needed. Read the result back with view data get. Evidence: docs/capability-evidence/ilg-sim-20260919.'
   reviewed_by: primary
 - command_id: view.transform.ai
   command_path: view transform ai