A plain string value (e.g. FAB_API_KEY_PREFIXES = "sst_") would iterate
as individual characters ['s','s','t','_'], matching far too many tokens.
Wrap strings in a list at the config-read boundary so CompositeTokenVerifier
always receives a proper sequence regardless of how the config is set.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Remove mock_sm.find_user_with_relationships.return_value = None from
_mock_sm_ctx: load_user_with_relationships delegates to the global
security_manager (not app.appbuilder.sm), so setting it on mock_sm had
no effect and broke MagicMock(spec=[]) tests.
- Add _patch_load_user_not_found() helper that patches
superset.mcp_service.auth.load_user_with_relationships directly.
- Apply it to the 3 JWT-path tests that expect ValueError("not found"):
test_jwt_access_token_skips_api_key_auth,
test_namespaced_claim_without_api_key_client_id_is_ignored,
test_unnamespaced_passthrough_claim_does_not_trigger_api_key_path.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Empty-string prefixes match every Bearer token (DoS/misclassification vector).
Non-string entries cause TypeError in str.startswith(). Filter both in __init__,
warn on invalid entries, and only store valid non-empty string prefixes.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
_mock_sm_ctx now sets find_user_with_relationships.return_value = None so
JWT/dev-user lookups that delegate through the (now refactored)
load_user_with_relationships → security_manager.find_user_with_relationships
path behave as "user not found" in unit tests that don't patch the DB — matching
the behavior of the previous direct db.session.query() implementation.
Without this, tests that expected ValueError("not found") received a truthy
MagicMock() from the unspecified mock method, causing them to fail.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Fixes a gap identified in code review: the standalone load_user_with_relationships()
in auth.py duplicated SecurityManager.find_user() logic but dropped two FAB behaviors:
- auth_username_ci (case-insensitive username lookup)
- MultipleResultsFound guard (username uniqueness not guaranteed at DB level in all FAB versions)
It also hard-coded User/Group models instead of sm.user_model.
Changes:
- Add SupersetSecurityManager.find_user_with_relationships() to security/manager.py,
mirroring FAB's find_user() (auth_username_ci, MultipleResultsFound handling,
self.user_model) and adding eager loading of roles and group.roles via joinedload
- Simplify load_user_with_relationships() in auth.py to a thin delegate to the
new method, removing the duplicated query logic and raw Group/User imports
- Add regression test asserting find_user_with_relationships() exists on the SM
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- _tool_allowed_for_current_user (server.py): catch PermissionError
alongside ValueError so invalid API keys return False instead of
propagating through the tool-search permission filter
- _setup_user_context (auth.py): catch PermissionError alongside
ValueError so g.user is cleared and the error is logged consistently
regardless of which failure type get_user_from_request() raises
- _resolve_user_from_api_key (auth.py): require client_id=="api_key"
(set by CompositeTokenVerifier) in addition to API_KEY_PASSTHROUGH_CLAIM
to prevent an external IdP JWT that happens to include the claim name
from being misclassified as an API-key pass-through (DoS vector)
- _resolve_user_from_jwt_context (auth.py): same client_id guard so
a rogue-claim JWT continues through JWT resolution instead of deferring
to the API-key path (which would raise PermissionError for the user)
- _resolve_user_from_api_key (auth.py): raise PermissionError (not
return None) when the pass-through claim is present but the raw token
is absent — fail closed rather than falling through to weaker auth
- Tests: set client_id="api_key" on _passthrough_access_token helper;
update test_jwt_context_with_api_key_passthrough_returns_none docstring;
add test for namespaced claim on non-API-key client_id being ignored
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Use superset.mcp_service.auth.has_request_context as patch target in
test_mcp_auth_hook_clears_stale_g_user tests; patching flask.has_request_context
has no effect on the module-level import already bound in auth.py
- Update test_jwt_access_token_skips_api_key_auth docstring to reference
API_KEY_PASSTHROUGH_CLAIM instead of the legacy _api_key_passthrough name
- Add noqa: BLE001 to broad exception catch in mcp_config.py to document
that the wide catch is intentional (JWT libs raise many types, secrets guard)
DetailedJWTVerifier and JWTVerifier have no circular-import or optional-
dependency reason to be imported inline — fastmcp is already pulled in
at module top via composite_token_verifier, and authlib is already a
hard dependency. Moving them up for consistency with the rest of the
module's imports.
``superset init`` calls ``appbuilder.add_permissions(update_perms=True)``
before ``sync_role_definitions()`` (cli/main.py:84), which forces FAB to
walk all registered baseviews — including ``ApiKeyApi`` (registered when
``FAB_API_KEY_ENABLED=True``) — and create their PVMs via
``add_permissions_view``. The explicit ``add_permission_view_menu`` calls
in ``create_custom_permissions`` were redundant.
With ``"ApiKey"`` already in ``ADMIN_ONLY_VIEW_MENUS``, the role
predicate ``_is_admin_only`` gates the auto-created PVMs to Admin.
Per Daniel Gaspar's review: "Adding ApiKey to ADMIN_ONLY_VIEW_MENUS
should just work when FAB_API_KEY_ENABLED is True".
The API_KEY_PASSTHROUGH_CLAIM constant in auth.py and CompositeTokenVerifier
in mcp_config.py have no circular-import or optional-dependency reason to
be imported inline. Moved them to module top.
Three independent bugs let MCP requests presenting Bearer tokens with the
sst_ prefix authenticate as MCP_DEV_USERNAME without any validation under
streamable-http:
1. _resolve_user_from_api_key read the token from flask.request.headers,
but the streamable-http transport never pushes a Flask request context
— has_request_context() was always False, so the function returned
None before validating, falling through to the dev-user fallback.
Now reads the token from FastMCP's per-request AccessToken (which the
CompositeTokenVerifier already populated) and fails closed when the
key is invalid.
2. CompositeTokenVerifier was only installed when MCP_AUTH_ENABLED=True.
With FAB_API_KEY_ENABLED=True alone, no transport-level verifier
existed at all. The factory now builds an API-key-only verifier in
that case (jwt_verifier=None) that rejects non-API-key Bearer tokens
at the transport instead of silently accepting them.
3. The pass-through AccessToken was minted with scopes=[], which would
make FastMCP's RequireAuthMiddleware 403 every API-key request when
MCP_REQUIRED_SCOPES is non-empty. Pass-through now propagates
self.required_scopes.
Also addresses Daniel's review comment on superset/security/manager.py:
adds "ApiKey" to ADMIN_ONLY_VIEW_MENUS so the FAB ApiKeyApi PVMs are
gated to Admin instead of leaking to Alpha and Gamma.
Renames the pass-through claim from _api_key_passthrough to the
namespaced _superset_mcp_api_key_passthrough (exported as
API_KEY_PASSTHROUGH_CLAIM) so a custom claim from an external IdP can't
accidentally divert a JWT into the API-key validation path.
Tests updated to mock get_access_token instead of app.test_request_context
(the simulated Flask context was the reason the prior tests passed while
production failed). New tests cover API-key-only verifier mode, scope
propagation on pass-through, and the namespaced-claim isolation.
Wire CompositeTokenVerifier into create_default_mcp_auth_factory,
add _api_key_passthrough detection in _resolve_user_from_jwt_context,
create ApiKey permissions in create_custom_permissions, and update
test_auth_api_key with pass-through and non-matching prefix tests.
Two fixes for MCP API key authentication:
1. superset init now creates ApiKey FAB permissions (can_list, can_create,
can_get, can_delete) when FAB_API_KEY_ENABLED=True. Previously, because
Superset uses AppBuilder(update_perms=False), FAB skipped permission
creation during blueprint registration and superset init never picked
them up, causing 403 errors on /api/v1/security/api_keys/.
2. CompositeTokenVerifier allows API key tokens (e.g. sst_...) to coexist
with JWT auth on the MCP transport layer. Previously, when
MCP_AUTH_ENABLED=True, the JWTVerifier rejected all non-JWT Bearer
tokens at the transport layer before they could reach the Flask-level
_resolve_user_from_api_key() handler. The composite verifier detects
API key prefixes and passes them through with a marker claim, letting
the existing auth priority chain handle validation.
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Joe Li <joe@preset.io>
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Joe Li <joe@preset.io>
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Joe Li <joe@preset.io>
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Joe Li <joe@preset.io>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: codeant-ai-for-open-source[bot] <244253245+codeant-ai-for-open-source[bot]@users.noreply.github.com>
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Joe Li <joe@preset.io>
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Joe Li <joe@preset.io>
We kindly ask you to include the following information in your report:
- Apache Superset version that you are using
- A sanitized copy of your `superset_config.py` file or any config overrides
-Detailed steps to reproduce the vulnerability
**Submission Standards & AI Policy**
To ensure engineering focus remains on verified risks and to manage high reporting volumes, all reports must meet the following criteria:
-Plain Text Format: In accordance with Apache guidelines, please provide all details in plain text within the email body. Avoid sending PDFs, Word documents, or password-protected archives.
- Mandatory AI Disclosure: If you utilized Large Language Models (LLMs) or AI tools to identify a flaw or assist in writing a report, you must disclose this in your submission so our triage team can contextualize the findings.
- Human-Verified PoC: All submissions must include a manual, step-by-step Proof of Concept (PoC) performed on a supported release. Raw AI outputs, hypothetical chat transcripts, or unverified scanner logs will be closed as Invalid.
We kindly ask you to include the following information in your report to assist our developers in triaging and remediating issues efficiently:
- Version/Commit: The specific version of Apache Superset or the Git commit hash you are using.
- Configuration: A sanitized copy of your `superset_config.py` file or any config overrides.
- Environment: Your deployment method (e.g., Docker Compose, Helm, or source) and relevant OS/Browser details.
- Impacted Component: Identification of the affected area (e.g., Python backend, React frontend, or a specific database connector).
- Expected vs. Actual Behavior: A clear description of the intended system behavior versus the observed vulnerability.
- Detailed Reproduction Steps: Clear, manual steps to reproduce the vulnerability.
**Out of Scope Vulnerabilities**
To prioritize engineering efforts on genuine architectural risks, the following scenarios are explicitly out of scope and will not be issued a CVE:
- Attacks requiring Admin privileges: (e.g., CSS injection, template manipulation, dashboard ownership overrides, or modifying global system settings). Per the CVE vulnerability definition in CNA Operational Rules 4.1, a qualifying vulnerability must allow violation of a security policy. The Admin role is a fully trusted operational boundary defined by Apache Superset's security policy; actions within this boundary do not violate that policy and are therefore considered intended capabilities 'by design,' not vulnerabilities.
- Brute Force and Rate Limiting: Reports targeting a lack of resource exhaustion protections, generic rate-limiting, or volumetric Denial of Service (DoS) attempts.
- Theoretical attack vectors: Issues without a demonstrable, reproducible exploit path.
- Non-Exploitable Findings: Missing security headers, generic banner disclosures, or descriptive error messages that do not lead to a direct, documented exploit.
**Outcome of Reports**
Reports that are deemed out-of-scope for a CVE but represent valid security best practices or hardening opportunities may be converted into public GitHub issues. This allows the community to contribute to the general hardening of the platform even when a specific vulnerability threshold is not met.
Note that Apache Superset is not responsible for any third-party dependencies that may
have security issues. Any vulnerabilities found in third-party dependencies should be
@@ -29,6 +51,13 @@ reported to the maintainers of those projects. Results from security scans of Ap
Superset dependencies found on its official Docker image can be remediated at release time
by extending the image itself.
**Vulnerability Aggregation & CVE Attribution**
In accordance with MITRE CNA Operational Rules (4.1.10, 4.1.11, and 4.2.13), Apache Superset issues CVEs based on the underlying architectural root cause rather than the number of affected endpoints or exploit payloads.
- Aggregation: If multiple exploit vectors stem from the same programmatic failure or shared vulnerable code, they must be aggregated into a single, comprehensive report.
- Independent Fixes: Separate CVEs will only be assigned if the vulnerabilities reside in decoupled architectural modules and can be fixed independently of one another.
Reports that fail to aggregate related findings will be merged during triage to ensure an accurate and defensible CVE record.
**Your responsible disclosure and collaboration are invaluable.**
When the feature flag is enabled, these permissions are enforced on both the frontend (disabled buttons with tooltips) and backend (403 responses from API endpoints). When disabled, legacy `can_csv` behavior is preserved.
**Migration behavior:** All three new permissions are granted to every role that currently has `can_csv`, preserving existing access. Admins can then selectively revoke individual export permissions from specific roles as needed.
### Deck.gl MapBox viewport and opacity controls are functional
The Deck.gl MapBox chart's **Opacity**, **Default longitude**, **Default latitude**, and **Zoom** controls were previously non-functional — changing them had no effect on the rendered map. These controls are now wired up correctly.
@@ -308,13 +322,13 @@ Note: Pillow is now a required dependency (previously optional) to support image
There's a migration added that can potentially affect a significant number of existing charts.
- [32317](https://github.com/apache/superset/pull/32317) The horizontal filter bar feature is now out of testing/beta development and its feature flag `HORIZONTAL_FILTER_BAR` has been removed.
- [31590](https://github.com/apache/superset/pull/31590) Marks the begining of intricate work around supporting dynamic Theming, and breaks support for [THEME_OVERRIDES](https://github.com/apache/superset/blob/732de4ac7fae88e29b7f123b6cbb2d7cd411b0e4/superset/config.py#L671) in favor of a new theming system based on AntD V5. Likely this will be in disrepair until settling over the 5.x lifecycle.
- [32432](https://github.com/apache/superset/pull/31260) Moves the List Roles FAB view to the frontend and requires `FAB_ADD_SECURITY_API` to be enabled in the configuration and `superset init` to be executed.
- [32432](https://github.com/apache/superset/pull/32432) Moves the List Roles FAB view to the frontend and requires `FAB_ADD_SECURITY_API` to be enabled in the configuration and `superset init` to be executed.
- [34319](https://github.com/apache/superset/pull/34319) Drill to Detail and Drill By is now supported in Embedded mode, and also with the `DASHBOARD_RBAC` FF. If you don't want to expose these features in Embedded / `DASHBOARD_RBAC`, make sure the roles used for Embedded / `DASHBOARD_RBAC`don't have the required permissions to perform D2D actions.
## 5.0.0
- [31976](https://github.com/apache/superset/pull/31976) Removed the `DISABLE_LEGACY_DATASOURCE_EDITOR` feature flag. The previous value of the feature flag was `True` and now the feature is permanently removed.
- [31959](https://github.com/apache/superset/pull/32000) Removes CSV_UPLOAD_MAX_SIZE config, use your web server to control file upload size.
- [32000](https://github.com/apache/superset/pull/32000) Removes CSV_UPLOAD_MAX_SIZE config, use your web server to control file upload size.
- [31959](https://github.com/apache/superset/pull/31959) Removes the following endpoints from data uploads: `/api/v1/database/<id>/<file type>_upload` and `/api/v1/database/<file type>_metadata`, in favour of new one (Details on the PR). And simplifies permissions.
- [31844](https://github.com/apache/superset/pull/31844) The `ALERT_REPORTS_EXECUTE_AS` and `THUMBNAILS_EXECUTE_AS` config parameters have been renamed to `ALERT_REPORTS_EXECUTORS` and `THUMBNAILS_EXECUTORS` respectively. A new config flag `CACHE_WARMUP_EXECUTORS` has also been introduced to be able to control which user is used to execute cache warmup tasks. Finally, the config flag `THUMBNAILS_SELENIUM_USER` has been removed. To use a fixed executor for async tasks, use the new `FixedExecutor` class. See the config and docs for more info on setting up different executor profiles.
- [31894](https://github.com/apache/superset/pull/31894) Domain sharding is deprecated in favor of HTTP2. The `SUPERSET_WEBSERVER_DOMAINS` configuration will be removed in the next major version (6.0)
Licensed to the Apache Software Foundation (ASF) under one
or more contributor license agreements. See the NOTICE file
distributed with this work for additional information
regarding copyright ownership. The ASF licenses this file
to you under the Apache License, Version 2.0 (the
"License"); you may not use this file except in compliance
with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing,
software distributed under the License is distributed on an
"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
KIND, either express or implied. See the License for the
specific language governing permissions and limitations
under the License.
*/}
---
title: AWS IAM Authentication
sidebar_label: AWS IAM Authentication
sidebar_position: 15
---
# AWS IAM Authentication for AWS Databases
Superset supports IAM-based authentication for **Amazon Aurora** (PostgreSQL and MySQL) and **Amazon Redshift**. IAM auth eliminates the need for database passwords — Superset generates a short-lived auth token using temporary AWS credentials instead.
Cross-account IAM role assumption via STS `AssumeRole` is supported, allowing a Superset deployment in one AWS account to connect to databases in a different account.
## Prerequisites
- Enable the `AWS_DATABASE_IAM_AUTH` feature flag in `superset_config.py`. IAM authentication is gated behind this flag; if it is disabled, connections using `aws_iam` fail with *"AWS IAM database authentication is not enabled."*
```python
FEATURE_FLAGS = {
"AWS_DATABASE_IAM_AUTH": True,
}
```
- `boto3` must be installed in your Superset environment:
```bash
pip install boto3
```
- The Superset server's IAM role (or static credentials) must have permission to call `sts:AssumeRole` (for cross-account) or the same-account permissions for the target service:
- **Redshift Serverless**: `redshift-serverless:GetCredentials` and `redshift-serverless:GetWorkgroup`
- SSL must be enabled on the Aurora / Redshift endpoint (required for IAM token auth).
## Configuration
IAM authentication is configured via the **encrypted_extra** field of the database connection. Access this field in the **Advanced** → **Security** section of the database connection form, under **Secure Extra**.
**3. Configure the database connection in Superset** using the `role_arn` and `external_id` from the trust policy (as shown in the configuration example above).
## Credential Caching
STS credentials are cached in memory keyed by `(role_arn, region, external_id)` with a 10-minute TTL. This reduces the number of STS API calls when multiple queries are executed with the same connection. Tokens are refreshed automatically before expiry.
The superset cli allows you to import and export datasources from and to YAML. Datasources include
databases. The data is expected to be organized in the following hierarchy:
:::info
Superset's ZIP-based import/export also covers **dashboards**, **charts**, and **saved queries**, exercised through the UI and REST API. The [Dashboard Import Overwrite Behavior](#dashboard-import-overwrite-behavior) and [UUIDs in API Responses](#uuids-in-api-responses) sections below document the behavior shared across all asset types.
:::
```text
├──databases
| ├──database_1
@@ -75,6 +79,29 @@ The optional username flag **-u** sets the user used for the datasource import.
When importing a dashboard ZIP with the **overwrite** option enabled, any existing charts that are part of the dashboard are **replaced** rather than duplicated. This applies to:
- Charts whose UUID matches a chart already present in the target instance
- The full chart configuration (query, visualization type, columns, metrics) is replaced by the imported version
If you import without the overwrite flag, existing charts with conflicting UUIDs are left unchanged and the import skips those objects. Use overwrite when you want to push a fully updated dashboard (including chart definitions) from a development or staging environment to production.
## UUIDs in API Responses
The REST API POST endpoints for **datasets**, **charts**, and **dashboards** include the auto-generated `uuid` field in the response body:
```json
{
"id": 42,
"uuid": "b8a8d5c3-1234-4abc-8def-0123456789ab",
...
}
```
UUIDs remain stable across import/export cycles and can be used for cross-environment workflows — for example, recording a UUID when creating a chart in development and using it to identify the matching chart after importing into production.
## Legacy Importing Datasources
### From older versions of Superset to current version
@@ -501,6 +501,7 @@ All MCP settings go in `superset_config.py`. Defaults are defined in `superset/m
| `MCP_SERVICE_URL` | `None` | Public base URL for MCP-generated links (set this when behind a reverse proxy) |
| `MCP_DEBUG` | `False` | Enable debug logging |
| `MCP_DEV_USERNAME` | -- | Superset username for development mode (no auth) |
| `MCP_PARSE_REQUEST_ENABLED` | `True` | Pre-parse MCP tool inputs from JSON strings into objects. Set to `False` for clients (Claude Desktop, LangChain) that do not double-serialize arguments — this produces cleaner tool schemas for those clients |
### Authentication
@@ -664,6 +665,32 @@ MCP_CSRF_CONFIG = {
---
## Audit Events
All MCP tool calls are logged to Superset's event logger, the same system used by the web UI (viewable at **Settings → Action Log**). Each event captures:
- **User**: the resolved Superset username from the JWT or dev config
- **Timestamp**: when the operation ran
This means MCP activity is auditable alongside normal user activity. No additional configuration is required — logging is on by default whenever the event logger is enabled in your Superset deployment.
## Tool Pagination
MCP list tools (`list_datasets`, `list_charts`, `list_dashboards`, `list_databases`) use **offset pagination** via `page` (1-based) and `page_size` parameters. Responses include `page`, `page_size`, `total_count`, `total_pages`, `has_previous`, and `has_next`. To iterate through all results:
```python
# Example: fetch all charts across pages
all_charts = []
page = 1
while True:
result = mcp.list_charts(page=page, page_size=50)
all_charts.extend(result["charts"])
if not result.get("has_next"):
break
page += 1
```
## Security Best Practices
- **Use TLS** for all production MCP endpoints -- place the server behind a reverse proxy with HTTPS
@@ -64,7 +64,7 @@ There are two approaches to making dashboards publicly accessible:
3. Edit each dashboard's properties and add the "Public" role
4. Only dashboards with the Public role explicitly assigned are visible to anonymous users
See the [Public role documentation](/admin-docs/security/security#public) for more details.
See the [Public role documentation](/admin-docs/security/#public) for more details.
#### Embedding a Public Dashboard
@@ -111,7 +111,7 @@ FEATURE_FLAGS = {
This flag only hides the logout button when Superset detects it is running inside an iframe. Users accessing Superset directly (not embedded) will still see the logout button regardless of this setting.
:::note
When embedding with SSO, also set `SESSION_COOKIE_SAMESITE = 'None'` and `SESSION_COOKIE_SECURE = True`. See [Security documentation](/docs/security/securing_superset) for details.
When embedding with SSO, also set `SESSION_COOKIE_SAMESITE = 'None'` and `SESSION_COOKIE_SECURE = True`. See [Security documentation](/admin-docs/security/securing_superset) for details.
@@ -22,6 +22,15 @@ While powerful, this feature executes template code on the server. Within the Su
If you grant these permissions to untrusted users, this feature can be exploited as a **Server-Side Template Injection (SSTI)** vulnerability. Do not enable `ENABLE_TEMPLATE_PROCESSING` unless you fully understand and accept the associated security risks.
Additionally:
- The `url_param()` macro allows URL parameters to influence the rendered SQL. Always validate or restrict `url_param()` values in your templates rather than interpolating them directly.
- `filter.get('val')` returns raw filter values without escaping. Use the safe helpers described below (`|where_in`, `| replace("'", "''")`) rather than concatenating values directly into SQL strings.
:::
:::tip
`ENABLE_TEMPLATE_PROCESSING` defaults to `False`. Only enable it if your deployment requires Jinja templates and all users with dataset/chart edit access are administrators or fully trusted internal users.
:::
When templating is enabled, python code can be embedded in virtual datasets and
@@ -324,6 +333,16 @@ cache hit in the future and Superset can retrieve cached data.
The `{{ url_param('custom_variable') }}` macro lets you define arbitrary URL
parameters and reference them in your SQL code.
:::warning
Always treat `url_param()` values as untrusted input. Escaping behaviour varies by context and configuration, so do not rely on it. Restrict values to an explicit allowlist before using them in SQL:
```sql
{% set cc = url_param('countrycode') %}
{% if cc not in ('US', 'ES', 'FR') %}{% set cc = 'US' %}{% endif %}
WHERE country_code = '{{ cc }}'
```
:::
Here's a concrete example:
- You write the following query in SQL Lab:
@@ -398,6 +417,16 @@ This is useful if:
- You want to handle generating custom SQL conditions for a filter
- You want to have the ability to filter inside the main query for speed purposes
:::warning
`filter.get('val')` returns the raw filter value without escaping. For multi-value filters, use the `|where_in` Jinja filter, which handles quoting safely. For single-value operators like `LIKE`, escape single quotes before interpolating:
```sql
{%- if filter.get('op') == 'LIKE' -%}
AND full_name LIKE '{{ filter.get('val') | replace("'", "''") }}'
{%- endif -%}
```
:::
Here's a concrete example:
```sql
@@ -424,7 +453,7 @@ Here's a concrete example:
{%- if filter.get('op') == 'LIKE' -%}
AND
full_name LIKE {{ "'" + filter.get('val') + "'" }}
full_name LIKE '{{ filter.get('val') | replace("'", "''") }}'
@@ -20,7 +20,7 @@ To help make the problem somewhat tractable—given that Apache Superset has no
To strive for data consistency (regardless of the timezone of the client) the Apache Superset backend tries to ensure that any timestamp sent to the client has an explicit (or semi-explicit as in the case with [Epoch time](https://en.wikipedia.org/wiki/Unix_time) which is always in reference to UTC) timezone encoded within.
The challenge however lies with the slew of [database engines](/admin-docs/databases#installing-drivers-in-docker) which Apache Superset supports and various inconsistencies between their [Python Database API (DB-API)](https://www.python.org/dev/peps/pep-0249/) implementations combined with the fact that we use [Pandas](https://pandas.pydata.org/) to read SQL into a DataFrame prior to serializing to JSON. Regrettably Pandas ignores the DB-API [type_code](https://www.python.org/dev/peps/pep-0249/#type-objects) relying by default on the underlying Python type returned by the DB-API. Currently only a subset of the supported database engines work correctly with Pandas, i.e., ensuring timestamps without an explicit timestamp are serializd to JSON with the server timezone, thus guaranteeing the client will display timestamps in a consistent manner irrespective of the client's timezone.
The challenge however lies with the slew of [database engines](/user-docs/databases#installing-drivers-in-docker) which Apache Superset supports and various inconsistencies between their [Python Database API (DB-API)](https://www.python.org/dev/peps/pep-0249/) implementations combined with the fact that we use [Pandas](https://pandas.pydata.org/) to read SQL into a DataFrame prior to serializing to JSON. Regrettably Pandas ignores the DB-API [type_code](https://www.python.org/dev/peps/pep-0249/#type-objects) relying by default on the underlying Python type returned by the DB-API. Currently only a subset of the supported database engines work correctly with Pandas, i.e., ensuring timestamps without an explicit timestamp are serialized to JSON with the server timezone, thus guaranteeing the client will display timestamps in a consistent manner irrespective of the client's timezone.
For example the following is a comparison of MySQL and Presto,
@@ -24,6 +24,14 @@ A table with the permissions for these roles can be found at [/RESOURCES/STANDAR
Admins have all possible rights, including granting or revoking rights from other
users and altering other people’s slices and dashboards.
>#### Threat Model and Privilege Boundaries: The Admin Role
>
>Apache Superset is built with a granular permission model where users assigned the Admin role are considered fully trusted. Admins possess complete control over the application's configuration, UI rendering, and access controls.
>
>Consequently, actions performed by an Admin that alter the application's behavior or presentation—such as injecting custom CSS, modifying Jinja templates, or altering security flags—are intended administrative capabilities by design.
>
>In accordance with MITRE CNA Rule 4.1, a vulnerability must represent a violation of an explicit security policy. Because the Admin role is defined as a trusted operational boundary, actions executed with Admin privileges do not cross a security perimeter. Therefore, exploit vectors that strictly require Admin access are not classified as security vulnerabilities and are ineligible for CVE assignment.
### Alpha
Alpha users have access to all data sources, but they cannot grant or revoke access
@@ -231,26 +239,143 @@ based on the roles and permissions that were attributed.
### Row Level Security
Using Row Level Security filters (under the **Security** menu) you can create filters
that are assigned to a particular table, as well as a set of roles.
that are assigned to a particular dataset, as well as a set of roles.
If you want members of the Finance team to only have access to
rows where `department = "finance"`, you could:
- Create a Row Level Security filter with that clause (`department = "finance"`)
- Then assign the clause to the **Finance** role and the table it applies to
- Then assign the clause to the **Finance** role and the dataset it applies to
The **clause** field, which can contain arbitrary text, is then added to the generated
SQL statement’s WHERE clause. So you could even do something like create a filter
SQL statement's WHERE clause. So you could even do something like create a filter
for the last 30 days and apply it to a specific role, with a clause
like `date_field > DATE_SUB(NOW(), INTERVAL 30 DAY)`. It can also support
multiple conditions: `client_id = 6` AND `advertiser="foo"`, etc.
All relevant Row level security filters will be combined together (under the hood,
the different SQL clauses are combined using AND statements). This means it's
possible to create a situation where two roles conflict in such a way as to limit a table subset to empty.
RLS clauses also support **Jinja templating** when `ENABLE_TEMPLATE_PROCESSING` is enabled, so you can write dynamic filters such as
`user_id = '{{ current_username() }}'` to restrict rows based on the logged-in user.
For example, the filters `client_id=4` and `client_id=5`, applied to a role,
will result in users of that role having `client_id=4` AND `client_id=5`
added to their query, which can never be true.
#### Filter Types
There are two types of RLS filters:
- **Regular** — The filter clause is applied when the querying user belongs to one of the
roles assigned to the filter. Use this to restrict what specific roles can see.
- **Base** — The filter clause is applied to **all** users _except_ those in the assigned
roles. Use this to define a default restriction that privileged roles (e.g. Admin) are
exempt from. For example, a Base filter with clause `1 = 0` and the Admin role would
hide all rows from everyone except Admin — useful as a deny-by-default baseline.
#### Group Keys and Filter Combination
All applicable RLS filters are combined before being added to the query. The combination
rules are:
- Filters that share the **same group key** are combined with **OR** (any match within
the group is sufficient).
- Different filter groups (different group keys, or no group key) are combined with
**AND** (all groups must match).
- Filters with **no group key** are each treated as their own group and are always AND'd.
| `POST` | [Create a new dashboard](/developer-docs/api/create-a-new-dashboard) | `/api/v1/dashboard/` |
| `GET` | [Get metadata information about this API resource (dashboard--info)](/developer-docs/api/get-metadata-information-about-this-api-resource-dashboard-info) | `/api/v1/dashboard/_info` |
| `POST` | [Create a copy of an existing dashboard](/developer-docs/api/create-a-copy-of-an-existing-dashboard) | `/api/v1/dashboard/{id_or_slug}/copy/` |
| `DELETE` | [Delete a dashboard](/developer-docs/api/delete-a-dashboard) | `/api/v1/dashboard/{pk}` |
| `PUT` | [Update a dashboard](/developer-docs/api/update-a-dashboard) | `/api/v1/dashboard/{pk}` |
| `POST` | [Compute and cache a screenshot (dashboard-pk-cache-dashboard-screenshot)](/developer-docs/api/compute-and-cache-a-screenshot-dashboard-pk-cache-dashboard-screenshot) | `/api/v1/dashboard/{pk}/cache_dashboard_screenshot/` |
| `PUT` | [Update chart customizations configuration for a dashboard.](/developer-docs/api/update-chart-customizations-configuration-for-a-dashboard) | `/api/v1/dashboard/{pk}/chart_customizations` |
| `PUT` | [Update colors configuration for a dashboard.](/developer-docs/api/update-colors-configuration-for-a-dashboard) | `/api/v1/dashboard/{pk}/colors` |
| `GET` | [Export dashboard as example bundle](/developer-docs/api/export-dashboard-as-example-bundle) | `/api/v1/dashboard/{pk}/export_as_example/` |
| `DELETE` | [Remove the dashboard from the user favorite list](/developer-docs/api/remove-the-dashboard-from-the-user-favorite-list) | `/api/v1/dashboard/{pk}/favorites/` |
| `POST` | [Mark the dashboard as favorite for the current user](/developer-docs/api/mark-the-dashboard-as-favorite-for-the-current-user) | `/api/v1/dashboard/{pk}/favorites/` |
| `PUT` | [Update native filters configuration for a dashboard.](/developer-docs/api/update-native-filters-configuration-for-a-dashboard) | `/api/v1/dashboard/{pk}/filters` |
| `GET` | [Get a computed screenshot from cache (dashboard-pk-screenshot-digest)](/developer-docs/api/get-a-computed-screenshot-from-cache-dashboard-pk-screenshot-digest) | `/api/v1/dashboard/{pk}/screenshot/{digest}/` |
| `GET` | [Get a list of charts](/developer-docs/api/get-a-list-of-charts) | `/api/v1/chart/` |
| `POST` | [Create a new chart](/developer-docs/api/create-a-new-chart) | `/api/v1/chart/` |
| `GET` | [Get metadata information about this API resource (chart--info)](/developer-docs/api/get-metadata-information-about-this-api-resource-chart-info) | `/api/v1/chart/_info` |
| `GET` | [Get a list of datasets](/developer-docs/api/get-a-list-of-datasets) | `/api/v1/dataset/` |
| `POST` | [Create a new dataset](/developer-docs/api/create-a-new-dataset) | `/api/v1/dataset/` |
| `GET` | [Get metadata information about this API resource (dataset--info)](/developer-docs/api/get-metadata-information-about-this-api-resource-dataset-info) | `/api/v1/dataset/_info` |
| `GET` | [Get a dataset](/developer-docs/api/get-a-dataset) | `/api/v1/dataset/{id_or_uuid}` |
| `GET` | [Get charts and dashboards count associated to a dataset](/developer-docs/api/get-charts-and-dashboards-count-associated-to-a-dataset) | `/api/v1/dataset/{id_or_uuid}/related_objects` |
| `DELETE` | [Delete a dataset](/developer-docs/api/delete-a-dataset) | `/api/v1/dataset/{pk}` |
| `GET` | [Get a dataset](/developer-docs/api/get-a-dataset) | `/api/v1/dataset/{pk}` |
| `PUT` | [Update a dataset](/developer-docs/api/update-a-dataset) | `/api/v1/dataset/{pk}` |
| `DELETE` | [Delete a dataset column](/developer-docs/api/delete-a-dataset-column) | `/api/v1/dataset/{pk}/column/{column_id}` |
| `DELETE` | [Delete a dataset metric](/developer-docs/api/delete-a-dataset-metric) | `/api/v1/dataset/{pk}/metric/{metric_id}` |
| `PUT` | [Refresh and update columns of a dataset](/developer-docs/api/refresh-and-update-columns-of-a-dataset) | `/api/v1/dataset/{pk}/refresh` |
| `GET` | [Get charts and dashboards count associated to a dataset](/developer-docs/api/get-charts-and-dashboards-count-associated-to-a-dataset) | `/api/v1/dataset/{pk}/related_objects` |
| `GET` | [Get distinct values from field data (dataset-distinct-column-name)](/developer-docs/api/get-distinct-values-from-field-data-dataset-distinct-column-name) | `/api/v1/dataset/distinct/{column_name}` |
| `POST` | [Duplicate a dataset](/developer-docs/api/duplicate-a-dataset) | `/api/v1/dataset/duplicate` |
| `GET` | [Get all schemas from a database](/developer-docs/api/get-all-schemas-from-a-database) | `/api/v1/database/{pk}/schemas/` |
| `GET` | [Get database select star for table (database-pk-select-star-table-name)](/developer-docs/api/get-database-select-star-for-table-database-pk-select-star-table-name) | `/api/v1/database/{pk}/select_star/{table_name}/` |
| `GET` | [Get database select star for table (database-pk-select-star-table-name-schema-name)](/developer-docs/api/get-database-select-star-for-table-database-pk-select-star-table-name-schema-name) | `/api/v1/database/{pk}/select_star/{table_name}/{schema_name}/` |
| `DELETE` | [Delete a SSH tunnel](/developer-docs/api/delete-a-ssh-tunnel) | `/api/v1/database/{pk}/ssh_tunnel/` |
| `POST` | [Re-sync all permissions for a database connection](/developer-docs/api/re-sync-all-permissions-for-a-database-connection) | `/api/v1/database/{pk}/sync_permissions/` |
| `GET` | [Get names of databases currently available](/developer-docs/api/get-names-of-databases-currently-available) | `/api/v1/database/available/` |
| `GET` | [Download database(s) and associated dataset(s) as a zip file](/developer-docs/api/download-database-s-and-associated-dataset-s-as-a-zip-file) | `/api/v1/database/export/` |
<summary><strong>Datasources</strong> (1 endpoints) — Query datasource metadata and column values.</summary>
<summary><strong>Datasources</strong> (2 endpoints) — Query datasource metadata and column values.</summary>
| Method | Endpoint | Description |
|--------|----------|-------------|
| `GET` | [Get possible values for a datasource column](/developer-docs/api/get-possible-values-for-a-datasource-column) | `/api/v1/datasource/{datasource_type}/{datasource_id}/column/{column_name}/values/` |
| `POST` | [Validate a SQL expression against a datasource](/developer-docs/api/validate-a-sql-expression-against-a-datasource) | `/api/v1/datasource/{datasource_type}/{datasource_id}/validate_expression/` |
</details>
<details>
<summary><strong>Advanced Data Type</strong> (2 endpoints) — Endpoints for advanced data type operations and conversions.</summary>
<summary><strong>Advanced Data Type</strong> (2 endpoints) — Advanced data type operations and conversions.</summary>
| Method | Endpoint | Description |
|--------|----------|-------------|
| `GET` | [Return an AdvancedDataTypeResponse](/developer-docs/api/return-an-advanceddatatyperesponse) | `/api/v1/advanced_data_type/convert` |
| `GET` | [Return an AdvancedDataTypeResponse](/developer-docs/api/return-an-advanced-data-type-response) | `/api/v1/advanced_data_type/convert` |
| `GET` | [Return a list of available advanced data types](/developer-docs/api/return-a-list-of-available-advanced-data-types) | `/api/v1/advanced_data_type/types` |
| `POST` | [Create a new permanent link (explore-permalink)](/developer-docs/api/create-a-new-permanent-link-explore-permalink) | `/api/v1/explore/permalink` |
| `POST` | [Create a new permanent link (sqllab-permalink)](/developer-docs/api/create-a-new-permanent-link-sqllab-permalink) | `/api/v1/sqllab/permalink` |
| `GET` | [Get permanent link state for SQLLab editor.](/developer-docs/api/get-permanent-link-state-for-sqllab-editor) | `/api/v1/sqllab/permalink/{key}` |
| `GET` | [Get permanent link state for SQLLab editor.](/developer-docs/api/get-permanent-link-state-for-sql-lab-editor) | `/api/v1/sqllab/permalink/{key}` |
| `GET` | [Get a list of themes](/developer-docs/api/get-a-list-of-themes) | `/api/v1/theme/` |
| `POST` | [Create a theme](/developer-docs/api/create-a-theme) | `/api/v1/theme/` |
| `GET` | [Get metadata information about this API resource (theme--info)](/developer-docs/api/get-metadata-information-about-this-api-resource-theme-info) | `/api/v1/theme/_info` |
| `DELETE` | [Delete a theme](/developer-docs/api/delete-a-theme) | `/api/v1/theme/{pk}` |
| `GET` | [Get a theme](/developer-docs/api/get-a-theme) | `/api/v1/theme/{pk}` |
| `PUT` | [Update a theme](/developer-docs/api/update-a-theme) | `/api/v1/theme/{pk}` |
| `PUT` | [Set a theme as the system dark theme](/developer-docs/api/set-a-theme-as-the-system-dark-theme) | `/api/v1/theme/{pk}/set_system_dark` |
| `PUT` | [Set a theme as the system default theme](/developer-docs/api/set-a-theme-as-the-system-default-theme) | `/api/v1/theme/{pk}/set_system_default` |
| `DELETE` | [Delete security user registrations by pk](/developer-docs/api/delete-security-user-registrations-by-pk) | `/api/v1/security/user_registrations/{pk}` |
| `GET` | [Get security user registrations by pk](/developer-docs/api/get-security-user-registrations-by-pk) | `/api/v1/security/user_registrations/{pk}` |
| `PUT` | [Update security user registrations by pk](/developer-docs/api/update-security-user-registrations-by-pk) | `/api/v1/security/user_registrations/{pk}` |
| `GET` | [Get distinct values from field data (security-user-registrations-distinct-column-name)](/developer-docs/api/get-distinct-values-from-field-data-security-user-registrations-distinct-column-name) | `/api/v1/security/user_registrations/distinct/{column_name}` |
| `GET` | [Get related fields data (security-user-registrations-related-column-name)](/developer-docs/api/get-related-fields-data-security-user-registrations-related-column-name) | `/api/v1/security/user_registrations/related/{column_name}` |
import { SafeMarkdown } from '@superset/components';
import { SafeMarkdown } from '@superset-ui/core/components';
```
---
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.