aminghadersohi opened a new pull request, #40959:
URL: https://github.com/apache/superset/pull/40959
### SUMMARY
Adds a new `duplicate_dashboard` MCP tool to the Superset MCP service. The
canonical AI workflow this enables: "create a regional/staging variant of this
dashboard" — clone a template dashboard, then edit the copy.
Design:
- Wraps the existing `CopyDashboardCommand` (the same backend used by the
frontend "Save as" flow on `/api/v1/dashboard/<id>/copy/`).
- The source dashboard can be identified by numeric ID, UUID, or slug
(resolved via `DashboardDAO.get_by_id_or_slug`, which applies dashboard access
filters).
- `DashboardCopySchema`'s copy data contract requires `json_metadata`; the
tool builds it server-side from the source dashboard's current state
(`json_metadata` plus a `positions` key from `position_json`), mirroring
exactly what the frontend "Save as" sends. `json_metadata` is intentionally not
exposed as a tool parameter.
- `duplicate_slices: bool = false` — when true, every chart on the source
dashboard is deep-copied into a new chart object owned by the caller; when
false the copy references the same charts.
- The new title is sanitized for XSS using the same pattern as
`generate_dashboard` (nh3-based `sanitize_user_input`), with a warning surfaced
when content is stripped and a hard error when the title sanitizes to nothing.
- Structured error responses for source-not-found, access-denied, forbidden
(`DASHBOARD_RBAC`), and copy failures; errors are wrapped for LLM-context
safety.
- Registered as a `mutate` tool (`class_permission_name="Dashboard"`,
`method_permission_name="write"`) and added to
`MCP_CACHE_CONFIG["excluded_tools"]` so responses are never cached.
### BEFORE/AFTER SCREENSHOTS OR ANIMATED GIF
N/A — MCP service tool, no UI changes.
### TESTING INSTRUCTIONS
1. Start the MCP service and connect an MCP client.
2. Call `duplicate_dashboard` with `{"dashboard_id": <existing
id|uuid|slug>, "dashboard_title": "My Copy"}` — verify a new dashboard is
created referencing the same charts, and the returned URL opens it.
3. Call again with `"duplicate_slices": true` — verify the new dashboard
contains new chart objects (different chart IDs) with the same layout.
4. Verify error responses for a non-existent dashboard ID and for a
dashboard the user cannot access.
5. Run unit tests: `pytest
tests/unit_tests/mcp_service/dashboard/tool/test_duplicate_dashboard.py`
### ADDITIONAL INFORMATION
- [ ] Has associated issue:
- [ ] Required feature flags:
- [ ] Changes UI
- [ ] Includes DB Migration (follow approval process in
[SIP-59](https://github.com/apache/superset/issues/13351))
- [ ] Migration is atomic, supports rollback & is backwards-compatible
- [ ] Confirm DB migration upgrade and downgrade tested
- [ ] Runtime estimates and downtime expectations provided
- [x] Introduces new feature or API
- [ ] Removes existing feature or API
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
