Community Discussions

Allow Custom Symbols for Dataset CategoriesWe would like the ability to assign custom icons or symbols to datasets rather than using the standard black dot that is displayed today.Allowing unique symbols would help users quickly distinguish between dataset types, sources, or business contexts, making the search results easier to scan and navigate. This would improve usability, particularly in environments with a large number of datasets.Expected BenefitsImproved discovery of newly added datasets Reduced user clicks and navigation friction Better visibility into available search content Faster identification of datasets through visual differentiation Improved overall search usability and adoptionThis enhancement would make the search experience more visually intuitive for end users. ​@Andrew Montgomery 

Search Experience ImprovementsWe would like to see the "Show More" expansion button removed from the Search experience.Today, as additional datasets are added, the interface progressively hides content behind the "Show More" control. This creates friction for users who must repeatedly click to access information that is already available. More importantly, users who are unaware that new datasets have been added may never discover them because they remain hidden by default.Instead, we recommend allowing the full list of datasets to be displayed in a scrollable view. If the number of datasets exceeds the available page space, users can simply scroll through the list rather than repeatedly expanding sections. This would improve discoverability, reduce clicks, and create a more intuitive search experience.Collapsed  Expanded​@Andrew Montgomery 

Hello Cognite team,First of all, thank you for your continuous development and improvements to CDF.I'd like to share one feature request based on our daily use of Charts and Canvas.Our use case In our long-term operation monitoring work, we create multiple Charts for different monitoring items and paste them onto a single Canvas, so that we can monitor many Charts together in one place.The issue The ability to select multiple Charts on Canvas and change their display period all at once is very convenient. However, when we open the same Canvas a month later and refresh to the latest data, the start time stays fixed at the point we originally set. As a result, the displayed time range keeps getting wider and wider over time, instead of showing a consistent window.What we'd like We would like a feature that lets us refresh Charts using the same fixed period — for example, a rolling window of the last 1 month, 3 months, or 1 year — so the time range stays consistent every time we open the

Hi team, I'd like to submit a feature request. Request In Charts, a line is only drawn up to the point where a value is registered in the time series. I'd like an option to draw a horizontal line from the last value up to the current time (NOW). This is especially an issue in calculations. When a formula uses multiple time series that are registered in CDF at different times, the series that ends earlier is plotted only up to its last value, so the calculation result is cut off there. I'd like to avoid this. ExampleGiven two time series Ts_long (up to today) and Ts_short (up to 2025/04/01), using them as calculation sources pulls the output toward the Ts_short time axis, so nothing is displayed after 2025/04/01.Reference テスト Copy（工務 久保田） | Cognite Data FusionCognite Hub

To keep clean structure in Charts monitoring jobs.

The current Cognite Grafana Plugin supports Time Series, Activities, and GraphQL-based queries, but does not provide native access to Records and Streams.As Records and Streams become increasingly important for data ingestion and operational use cases, native support within the Grafana Plugin would enable a more complete and consistent visualization experience directly in Grafana.Requested enhancement:Native querying of Records and Streams from the Grafana PluginSupport for filtering and visualization of Records dataIntegration with Grafana panels and dashboards without requiring intermediary services or custom pluginsThis enhancement would simplify data consumption, reduce implementation complexity, and improve the usability of Records and Streams across operational monitoring and reporting scenarios.

Using Search, when selecting an instance and then choosing the Properties tab, all the populated properties are listed with the key and value in two columns. The key column is apparently a fixed width column (40ish char wide), not possible to manually resize, and it doesn’t resize when the browser window is resized. This becomes a problem when the property key is long, something that may happen when the property is a JSONObject such that the key becomes “nested” Below is an example of how it looks It seems like the UI is able to wrap the line if the key either has a space or a hyphen, but not else (have not tried all characters...) We can of course mitigate this by creating more compact keys, but it would be really helpful if we could resize the column such that our key naming convention do not have to take a 40 character UI limitation into account. 

Hi Team,In regards to Cognite Data Model UI we understand that the below features are currently not supported and would like to add them here as a Product Idea requests -Reorder the Sequence of the Views listed in Cognite Data Model UI. For example, Under "Assets" category we have 1) Piping Tag 2) SAP Assets. We would want the "SAP Assets" View to come first followed by the "Piping Tag" View. This would make easier for end users by having important and frequently visited views on top. Also, would reduce the number of clicks a users has to make to get the the required data - for example in Canvas when we want to "Add Data", the default category is set to "Piping Tag" and user has to switch to "SAP Assets" to get to the right view.   Option to HIDE some of the views from the Data model UI main/home/landing page. Most of the Views need not be accessible from the Data Model main page and can be accessed through relationships. For example - Under "Activities" category we currently have 11

RequestTo improve business continuity and operational stability, we would like to request your consideration for developing official applications within InField that provide the following capabilities:An application that can export individual inspection forms in both CSV and PDF formats. An application that enables approval of individual inspection forms and allows users to view and track their approval status.BackgroundCurrently, we operate a workflow in which inspection results are entered through InField, exported using a Streamlit application, and then submitted for supervisor approval through another Streamlit application.Under normal circumstances, this workflow functions without major issues. However, due to a Streamlit application issue that occurred yesterday, the workflow became temporarily unavailable. To the best of our knowledge, this is the second occurrence of a similar issue this year.In addition, since Streamlit is currently provided as a preview feature, we have conce

Recently, the use of Charts has been increasing, and more users are creating complex calculations for data visualization.In this context, we have received feedback requesting a “Back” button (undo function).When users are building complex calculation formulas, there have been multiple instances where the calculation source suddenly disappears, forcing them to restart from scratch. The more complex the calculation, the greater the impact of this issue.To mitigate this, it would be extremely helpful to have a function that allows users to go back one step (e.g., an undo or back button).We believe this feature would significantly improve usability and reduce the risk of losing work during calculation editing. Thank you for your consideration.

Hi team,I'm a Cognite Data Fusion (CDF) user, and I'd like to submit a feature request regarding scheduled calculations in Charts that write results to a time series tag.Current behavior (as I understand it) When you set up a scheduled calculation in Charts to store results as a time series, the calculation only runs from the point the schedule is configured onward. It does not compute results for any period before the setup time. I confirmed this behavior in the following Hub post: 🔗 Cognite HubWhat I'd like to see I'd love an option to backfill / calculate historical data retroactively, so that the scheduled calculation can also produce results for the period before the calculation was created.Why this matters With this capability, we would be able to go back and review historical data as well, not just data going forward. This would be very valuable for our long-term monitoring and trend analysis, where being able to look back over past periods is essential.Thank you very much for

Hi there, We would like to be able to annotate instances that do not extend CogniteAsset, nor CogniteFile and visualize that in “Preview” in Search UI. Our context is purely data modeling.So we would like to have a view, that extends CogniteDescriable and to be able to annotate it towards a CogniteFile.Currently we are achieving partically that, by having an edge, (pseudo code below) ANNOTATION_TYPE = "diagrams.InstanceLink" type_node = NodeApply(space=REFERENCE_SPACE, external_id=ANNOTATION_TYPE)edge = EdgeApply( space=EDGE_INSTANCE_SPACE, external_id="annotation_<file_ext_id>__<entity_ext_id>", type=DirectRelationReference(space=REFERENCE_SPACE, external_id=ANNOTATION_TYPE), start_node=DirectRelationReference(space=FILE_SPACE, external_id=FILE_EXTERNAL_ID), end_node=DirectRelationReference(space=ENTITY_SPACE, external_id=ENTITY_EXTERNAL_ID), sources=[ NodeOrEdgeData( source=ViewId("cdf_cdm", "CogniteDiagramAnnotation", "v1"),

**Component**: OPC-UA extractor · `EUInformation` handling **Current behaviour** (confirmed by Cognite): the extractor never reads `NamespaceUri` or`UnitId` from `EUInformation`. The only value that reaches any config surface is thecomposed `DisplayName: Description` string. The UNECE code is dropped before`metadata-mapping` can see it, so `unit` / `unitExternalId` resolution to a`CogniteUnit` from the source unit is impossible today and is not planned. **Request**: carry `UnitId` through, and resolve the CDF `unit` direct relation from it. **Why it matters to us**: across our fleet the `unit` direct relation is populated on**0 of 2772** time series, while 528 of 854 at one line carry the composed string in`sourceUnit`. The only path available is post-processing that string-matches against theCDF unit catalogue. That is brittle in exactly the way a code-keyed lookup is not —`kW·h: kilowatt hour` has to be parsed and matched by display text, and it will differby server locale, by punctu

So, when I get documents to populate in canvas, it drops them off in an unorganized bundle. I tried to get ATLAS to organize them in a timeline but ATLAS states “I can add documents to the current canvas, but I cannot move or rearrange existing canvas elements with the available tools. Please drag the four cards into this left-to-right order:”. Is an update that can make it easier to organized documents in canvas. It gets very cluttered and time consuming to organize these documents when you have like 20-30 documents populating on canvas.

As per the 3D Config feature, it should be good to have a Canvas Config option to configure the colouring for a quick highlight of the contextualization results in a Canvas for P&ID or any “plan” documents like isometrics loaded in a Canvas.In the configuration option we should be able to select attributes/data linked to the OCR captures, then in Canvas we should have a button to select and highlight all the OCR captures contextualized to PI-Tags, 3D tags, WOrkorders, SECE equipments…This option is of importance, it will avoid to check one by one each equipment in a P&ID, it will accelerate the completion of the contextualizations workflows, but not only… it’s also of importance for operations when preparing an intervention.

Due to the limit on the number or Views/revisions, we need to deprecate and delete old revisions on a regular basis (in periods with heavy data model development). For Atlas agents and Canvas’ we can find the View versions via the API, allowing us to identify users depending on models that are deprecated. However, with Charts I cannot find any way of programmatically finding this.Is there a hidden API somewhere that allows me to inspect what the different Charts are relying on? If not,. this is an important functionality for governance of data models and needs to be added. 

When creating calculations in CHARTS it would be VERY useful to have logic functions (AND, OR, NOT, XOR) which allow multiple (>6 for AND and OR) boolean inputs (TRUE/FALSE) and a single output.This would be really useful for tracking events which may be dependent on multiple conditions being active (or not active). 

If you in Charts add a timeseries that is contextualised to a CogniteAsset, you will get the “P&ID” link icon. By clicking on it you will enter search mode for that specific asset, showing the File links as default. This is a nice workaround for a “Find related data”, also allowing you to add CogniteActivities.The instance is shown as a CogniteAsset, and it obviously use the inspection endpoint since you are allowed to change between different views for this instance. However, it doesn’t show the View that is the CogniteAsset extension. The consequence of this is that I’m not able to see neither CogniteFiles nor CogniteActivities as their extension (my CogniteAsset extension has reverse direct relation to the CogniteFile and CogniteActivity extensions). This seriously limits me when filtering on important properties since they only exist in the extension.  A workaround is to go directly to “+ Add data → Add time series”, and then select my CogniteAsset extension when I search, but

When asking the embedded Atlas AI Agent to eg show the difference between two timeseries, it display both timeseries in a chart inside the Chat window. However, both lines get the same color, and there is no legend, so unless you know the average value for those timeseries you do not know which is which. Also, there is no y-axis, so the only way of knowing the value is to hover over the different lines.Legend and axis would help a lot on the user-friendliness  

Currently, monitoring alert emails cannot be customized beyond a certain point: the email subject is always “An alert was created” and the body of the email cannot be customized to be more user-friendly for readers, etc. I would like more customization for the alert emails.

Example: The filter shows options in English, while in the list view they are displayed in Norwegian. For example, the filter shows “Ready”, whereas in the view it shows “Klar”.

Documents that are most important for the technician should be listed first, e.g. P&IDs (XB). 

Hello,With the current WITSML extractor configuration, we can only whitelist the mnemonics that we want to extract. Would it be possible to also support a blacklist configuration for mnemonics?In our case, the complete list of mnemonics to extract is not known beforehand, which makes maintaining a whitelist challenging.Thank you. 

When listing instances of a view and using the to_pandas method, is there a simple way to use the data types defined on the container properties and use these to cast to similar python data types? Or could this be an option of the python SDK?In the classical CDF data types, this is implemented so that for example a date field and timestamp fields have the correct python data type. The data modelling instances will be either floats or string. It would be very convenient if the underlying container data type can be used so that you do not have to manually cast values in the pandas dataframe.

ContextWe currently run several custom Python extractors built on cognite-extractor-utils. For those, cognite-extractor-utils lets us configure a metrics.cognite block that periodically pushes the extractor's Prometheusmetrics (Counter/Gauge) directly to CDF Time Series — this gives us consistent,queryable observability (row counts, run duration, throughput, failure counts, etc.)across all our extractors, without any extra tooling.ProblemThe PI Extractor and the DB Extractor (official Cognite binaries) already exposePrometheus metrics internally (e.g. pi_extractor_timeseries_states,pi_extractor_stream_iterations, pi_extractor_pi_data_pipe_events,pi_extractor_streamer_data_points, pi_extractor_event_queue_size,pi_extractor_pi_data_loss, pi_extractor_pi_connection_time,pi_extractor_pi_connections, extractor_utils_cdf_datapoints, …), but today they canonly be exposed via a local Prometheus scrape endpoint (metrics.server) or a PushGateway (metrics.push-gateways). There is no equivalent of