We would like to be able to generate dashboards to be embedded in customer pages (on our website), when the dashboard is given the customer ID as a global parameter. Customers should never see other customers’ data.

To do this effectively we require the following features:

  • That the public URL can be generated with pre-set global parameters encoded in the URL

  • That these can be generated from an API call programatically (i.e sent the dashboard ID and parameter value, return the encoded public URL).

  • That the page linked to by this URL does not have the global parameters visible or accessible to change (this could be an option sent in the API call).

I’m familiar with Python (and a little with Flask) and quite happy to help implement these (at least at the API level for now, they needn’t appear in the Web page option for generating the public dashboard URL right away).

How are the public dashboard URLs generated at the moment? Is it hashed and then stored in the database as a hash : dashboard lookup, or is it encrypted and then decrypted by the server on the fly to lookup the dashboard?

Also, is this process applied to the URL/dashboard ID + parameters or is the whole dashboard seralised and then hashed or encrypted (in the former case, we wouldn’t need to regenerate the public URLs if we change a graph on the dashboard for example).

Many thanks,
James McMurray


Hi @jamesmcm,

Thank you for starting this discussion.

Currently parameters work by using the query text as a template and merging the parameters values into it when running the query. This is not safe and prone to SQL injections (by definition). For this reason we don’t support parameters in embeds or shared dashboards.

The use case you describe is a bit different, as you want to preset the parameters values, which makes the threat of SQL injections irrelevant as the person who assigns the parameter values is a trusted person who has access to the data.

The way share dashboards work is that we create an API Key object that has access to a given dashboard:

So this is how we currently generate a shared dashboard URL.

To bind a shared dashboard URL with a list of parameters we have two options (I can think of):

  1. We can store along with the API Key a list of parameter values that are allowed to be used with this API Key.
  2. Have the concept of a public API key and a private one. The public API key will be used to reference the dashboard and as a way of creating a random URL. The private API key will be known only to the system and will be used to sign the URL with a specific set of parameters.

The benefit of #1 is its simplicity and security (no way for an outside user to alter the list of values). The benefit of #2 is that we don’t have to create a database record for each set of parameters.

The next issue is how to load the data. Currently the public dashboard API returns the dashboard data along with the query results in one API call. To be able to run queries with parameters and utilize current APIs it’s best if we implement support in current Query Results API to deduce whether the current call is allowed based on the dashboard API key = if the dashboard has this query & the parameters are in the allowed list.

That’s a lot of text :slight_smile: I hope it all makes sense, please let me know if you have further questions and what are your thoughts.


Thanks! Option 1 would work, so we could generate these URLs with fixed parameter settings via the API, and attempting to change those from the dashboard would result in an error (but they would still be visible in the UI) ?

That would be perfect, the only other thing is that we might want some separate filters on the dashboard, i.e. for date ranges, with default values set that can be modified by the anonymous public user. Would that be possible in combination with your proposal?


I think that preset parameters won’t be rendered, so the user won’t even have the option to change them (or know they exist).

Eventually we want this to be possible, but that requires a more complete solution to the issue of allowing safely running parameterized queries by non trusted users.


What about escaping the input, so no SQL statements could get passed? Here’s some additional ideas: https://www.owasp.org/index.php/SQL_Injection_Prevention_Cheat_Sheet

Another idea would be to provide a regex filter for allowed params. That way you could have a date field , email address field, domain name field, etc. that gets checked against a whitelist regex filter. If it matches, its passed through, otherwise its rejected.

  • Ciaran
1 Like


what an interesting discussion.

We really appreciate readash search and query capabilities.
I work for an environmental NGO and I use Redash to create dashboards related to physicochemical parameters of French Mediterranean lagoons (about 100) from 2010 to now.

I need to share my dashboard with 40 people, with only their own lagoons and ponds.
Actually I can’t do that because even i I can, as connected user, set parameters using dashboard widget, when I share the dashboard by url, the final reader will see another result…

So I think this would be a great feature.
As the possibility to support parameters in embeds or shared dashboards !

1 Like

This puts the responsibility on the user, which in many cases might not be fully aware of the possible risks.

But regardless of the security aspect adding the option to validate the input content (along with some presets for common patterns) is a good idea.

Why not create them “proper” users instead of giving them shared dashboards? I would assume that you wouldn’t mind that they see data about the other ponds/lagoons, or would you?

In fact it is a “problem”, not a huge one but a problem. I agree they could connect to the server, choose the dashboard and filter but it is not as easy to use" as it could be and by the way they don’t really want to see all their data.
Another way should consist to use a “connected_user_email” parameter in the query tool so the data would be filter.
Thanks for your attention and thanks for the work !

1 Like

Wanted to add a “+1” to this conversation and confirm that the scenario of selecting from set of predefined parameters would also work for our scenario: dashboards that are the same reports but varying by location, integrating the visualization into a product flow.

This ability would make redash particularly suitable for those of us that want an unopinioned BI/Reporting system that we can use as a data visualization extension of our products.

However, we do not have quite as strong a security requirement. For us this is just a convenience method for presenting slices of public-ish data.

We could make either option 1 or option 2 work (i.e. we could make a call with a private key to generate a secure token that could be used in a “signature check of the parameter” assuming that is what you meant). Proposal 2 sounds to me like s3 presigned urls conceptually: https://docs.aws.amazon.com/AmazonS3/latest/dev/ShareObjectPreSignedURL.html (Which reminds me you could probably include an expiry timestamp if you want a concept of a “session duration” )

FWIW: I feel that, for many users, you could also make the case that it was appropriate due diligence from a security standpoint to just sanitize everything that comes in over the query string. This would protect the data source itself, and, would just require the person generating the dashboard share string to have a some understanding of the data. To deal with the original scenario of customers not seeing each others data the data itself could use “crypto unique” or “unguessable” id space for the customer ids. Which I think is almost equivalent because the proposals above still produce a URL that is “valid forever” (unless you add the timestamp). So maybe separate the concerns of sql injection from the concerns of access authorization.


Hello everyone,

We are currently looking for the same kind of solution for our redash.

Can you tell me if it is possible now or if there are other ideas we can try ?

Thanks everyone !

In terms of alternatives: we just did a really thin API on the data in python and created a web page to do charting. So not using redash, but, not too hard to set up either.

In terms of charting:

We have ended up using: http://jerairrest.github.io/react-chartjs-2/

But also were using this: https://nivo.rocks

I honestly prefer Nivo from a visualization standpoint

very interesting

what is the status of this request?

Bump. I’d like to know if there’s been any movement here. I’m looking for a way to give date-range control within the shared dashboard. We’re embedding the dashboards within iframes inside our own portal, and the other limitations of shared dashboards (no ability to view query, etc.) are actually very useful

My question answered by the V8 announcements. This is implemented! Huzzah!

@jamesmcm does V8 Beta cut the mustard for you?

1 Like

Hi @arikfr @chrismerck

Any suggestion on how to implement the key aspect of this request posed by James;

to allow for a hidden/locked parameter to be embedded in the url/token.

Use case:
We have 10 customers.
We have a SQL query where we filter customers so they only see their data…
We want to have generic query and have the customer field be populated via a parameter so that we dont have duplicate queries for each customer.
We want to embed the same dashboard for all 10 customer.
We dont want the user to be able to change the customer field/parameter. Technically customer parameter/field shouldn’t be visible to the user. But they can change other embedded parameters for the queries such as the date.
Chart IO has a similar feature where they embed the customer specific parameter in a JWT unique for each customer.


1 Like

This feature isn’t available yet. But the team is now actively developing it. Keep an eye on the announcements in the forum early next year for when this becomes available.


I have the same problem as barkev today. I would like to know if this feature is now available.

It is not available yet. We will announce on the forum when it is released.

Thank you. I look forward to it.

1 Like