Recently active
ProblemSmart Factory is a Flows custom app that runs inside the CDF iframe with hideShell=true. In that mode users cannot open the Fusion account menu to sign out.HostAppAPI from connectToHostApp (@cognite/app-sdk) already exposes getAccessToken(), getProject(), getBaseUrl(), navigateExternal, and setHideShell (see Flows Auth API). It does not provide a way to end the Cognite Auth / Fusion session from the app:There is no logout() (or similar) on HostAppAPI. Cognite Auth RP-initiated logout (end_session_endpoint at https://auth.cognite.com/oauth2/sessions/logout) expects an id_token_hint. The host only exposes a CDF access token via getAccessToken(), not an ID token. Fusion’s own logout uses that end-session endpoint with id_token_hint and post_logout_redirect_uri. From the iframe we cannot obtain a valid id_token_hint.We also cannot use setHideShell(false) just to expose Cognite’s logout UI. Proposed solutionExtend HostAppAPI with one of:logout(options?) – the host clears the Cognite
Hello Cognite Community,We are evaluating the throughput limits of the Cognite Hosted MQTT Extractor.Our use case involves subscribing to MQTT topics and storing the received messages in CDF RAW. Each MQTT message has a payload smaller than 1 KB.We would like to clarify the following: Are there any hard limits on MQTT message throughput (messages/second)? Per MQTT client connection Per subscription (topic filter) Per Hosted Extractor job If there are no fixed limits, what is the recommended maximum sustained throughput (messages/second)? What happens if the incoming message rate exceeds the Hosted MQTT Extractor's processing capacity? Are messages buffered? Are failed messages retried? Is there a risk of message loss? Do the CDF RAW API write quotas also apply to Hosted MQTT Extractors, or are there separate internal limits? We would appreciate any official documentation, benchmarks, or practical experience regarding these limitations.Thank you!
SummaryLet Atlas create a Chart calculation from a natural-language request and plot the result as a time series chart.Today a user can build that calculation by hand in the Charts node editor (sources, constants, functions, and an output node), or ask Atlas to analyze stored time series. Atlas can embed a bar chart, KPI, table, or an existing Charts time series in a reply. It cannot turn a formula, for example a pressure difference, a unit conversion, or a scaled flow, into a calculation and show that calculated series on a line chart unless the result already exists as a time series in CDF. ProblemEngineers and operators compare derived signals as often as raw tags: pressure drop, API gravity, unit conversion, ratio of two flows, a moving average. Those series are usually not stored in CDF.The path today is manual: Find the source time series. Open Flows > Charts, create a chart, and add the series. Open Add data > Create calculation and wire sources, constants, and functio
The ability to define a CogniteUnit on a column in a container is very powerful. However, there are many situations where we are not able to define one common unit for an entire column.One example is LCI/Equipment data. Properties like length, volume, weight will have different units for a massive compressor vs a tiny bolt. We want to avoid doing unit conversion on ingestion since it introduces added complexity and risk (how to handle that a unit is not specified in the source), and it will also require the end user to do the unit conversion to get the number on the unit from the source system. One possible approach could be to have a column with the CogniteUnit instanceid, and then instead of defining the fixed CogniteUnit for a column we define the column with the unit.
One of the more important properties for industrial data is the Engineering units of a value (unit of measure).Today we can add a CogniteUnit, both to a timeseries and to a column in a data model container, which will enable unit conversion on consumption.When we ingest the data from our source systems, we need to identify the unit as it is described in the source system, and then match it to the correct CogniteUnit. Many of our source systems do unfortunately not follow a standard for engineering units (leading to the need for the looooong alias list in the unit catalogue).But some of our systems specify the unit via a standard. For AkerBP it is mainly CFIHOS and OPC-UA (which really is UN/CEFACT unit codes)It would be very helpful to both being able to identify a unit via these standards when the unit is populated, but also when it is consumed.Eg: Temperature in Celsius isCFIHOS-60001435 CE/CEFACT/OPCUA unitID=4408652I can add the CFIHOS code to the alias list (we are doing this now)
Hello,I started doing https://learn.cognite.com/cognite-data-workflows/2026033 course but found that in cognite-learn organization there is no access for Data worflows enabled. I also checked the admin group for learners at it seems that such access is missing. Im attaching screenshots.I would be grateful for any guidance on this issue. And can I be granted such access?BestTom
Hello, Im doing https://learn.cognite.com/cognite-functions/1327986 Functions course and looks like some materials are outdated and some parts of the notebooks dont run. In the course materials there is information that new course is under development, but I couldnt find any new course on this topic.So is there a new one or this is the only course on Cognite Functions?Best
Following the training instructions(Date Engineer - Hands-on: Setting up the Equipment Container for RMDM), I installed cognite-toolkit==0.6.87. However, the training repository is pinned to:[modules]version = "0.6.19"This causes cdf build to fail with:"The version of the modules (0.6.19) does not match the version of the installed CLI (0.6.87)"The repo notes that the modules version should not be changed manually, so we're forced to downgrade the Toolkit, but this is not mentioned in the training material.
How should data write-back be implemented in an application developed using Cognite Flows? Specifically, where should application-generated files and structured data be persisted, which CDF services should be used for different data types, and how should the write-back layer integrate with the broader CDF architecture?
Hi Cognite Community,We are looking at ingesting approximately 1 million engineering and operational documents into CDF, including equipment manuals, datasheets, inspection reports, SIL/HAZOP studies, operating procedures, technical studies, maintenance reports, and other asset-related documents.Our key requirement is for an Atlas AI agent to answer questions across this entire document repository without the user having to identify or pinpoint a specific document first.For example, a user should be able to ask:“What are the recurring compressor seal issues reported across our engineering and operational documents?”The user may not know which documents contain the relevant information. We therefore expect Atlas AI to discover the appropriate documents/passages, retrieve the relevant content, and generate a grounded response with references to the source documents.According to Cognite's documentation, CDF Semantic Search does create vector embeddings for PDF documents as part of the bui
Several Data model views contain large volumes of historical data, which can impact performance, scalability, and overall user experience. A potential improvement would be to implement a historical data archiving strategy by separating older records from active datasets.For example, records beyond a defined retention period could be moved to dedicated archive views, while primary views retain only the most recent and frequently accessed data. Another option would be to maintain a separate historical reporting layer specifically designed for long-term analysis and archival reporting.This approach would help improve query performance, reduce dataset sizes, and enhance usability for analytics and reporting tools. Users would benefit from faster data retrieval, shorter dashboard load times, and a more responsive self-service analytics experience, while also reducing processing overhead on the platform.
Today we’re proud to announce the 2026 Class of Cognite MVPs: ten community members who have gone above and beyond in the Cognite Hub this past year.MVPs are the people who make this community what it is. They answer questions, share what they’ve learned from real projects, give honest feedback that helps shape our products, and welcome newcomers. Their expertise and generosity have helped many members get unstuck and get more value out of Cognite Data Fusion.Please join us in congratulating:Aditya Kotiyal, SLB André Alves, Radix Christophe Allanic, TotalEnergies Gunnar Andreas Staff, Aker BP Mami Kubota, Cosmo Oil Markus Pettersen, Aker BP Marwen Taleb, TotalEnergies Neerajkumar Bhatewara, SLB Pierre Rambourg, MaMaTa Shun Takase, Nippon ShokubaiTo everyone on this list: thank you! Your contributions make this community a better place for everyone, and we’re lucky to have you.Want to be part of next year’s class?MVPs are recognized for helpful answers, knowledge sharing, and the posit
Whether CDF natively supports semantic/ontology-based relationship modelling (named, typed relationships like contains, hasTimeSeries, performedOn with direction/cardinality/metadata), and whether Atlas AI and other consumers can traverse that semantic meaning rather than just node-to-node links.
In order to delete nodes in CDF in a given View because of deletes in the Fabric source table, we need to extract necessary fields for all rows in Fabric to construct the space and ExternalId, and then compare this with a full extraction of the space and externalId in the View.In one example with ~400k rows/nodes, it takes 1-2 minutes to get the data from Fabric. However, the cdf_nodes is several orders of magnutide slower.Only extracting space and externalId for 400k nodes in the View takes ~25 minutes Adding a Where clause to check if the externalId has a given prefix as a safeguard makes the query time out.Using cdf_data_models or other functions does not seem to help either. Is there a reason that listing ids for nodes is so slow? And are there other ways to go about this use case?
Hello, Is timeSeriesAcl:read acl required for data point subscription consumption? Although the client has timeSeriesSubscriptionsAcl:read. We recieve: CogniteAPIError: Missing capability to read Time series | code: 403. When calling: ‘client.time_series.subscriptions.iterate_data’.Thanks
Improve how the AI explains its answer to the final user.Today, users may receive a result without clearly understanding how the AI reached that conclusion or which information was considered.The proposed improvement is to provide a simple and user-friendly explanation that shows: What information was used to generate the answer. The main steps the AI followed. Progress while the answer is being prepared. A short explanation of why the AI reached that result. References or sources used, when available. The explanation should use simple, user-friendly language and avoid technical terms. It should also be provided in real time, using a streaming experience, so the user can follow the main reasoning steps while the answer is being generated instead of waiting until the end.This would help users better understand the answer, increase trust in the AI, and make it easier to validate the result.
Hi, Is it possible to perhaps add a "warning" status to Extraction Pipelines? Sometimes it is nice to not just flag something as a failure but still something that should be flagged/raised.
I feel that when zooming in with the mouse wheel, the increment at which it zooms is too small. For instance, when I zoom into a canvas with 5 drawings it takes 15 mouse wheel up’s to get to the zoom level that I am able to read any one of those drawings. Then when zooming back out it takes 15 to see the whole canvas again.
Hello Community, I am just trying to understand the zero copy connection to databricks. I understand from the examples at cognite-databricks/README.md at main · cognitedata/cognite-databricks · GitHub, that CDF exposed its data through the Unity Catalog, help register its views (including Core CDM etc) and query them from databricks notebook. How about the reverse - for example to access data from databricks from outside databricks, does “zero copy” applies. For example, I see one can query the SQL warehouse using service principals and PATs. Is there a zero copy way, that I am missing? ThanksAlex
RequestProvide the ability to change bounding box colors based on the status or metadata of the associated document, inspection data, or record.The goal is to make important status information visible directly within the working view, allowing users to recognize items that may require attention without opening each item individually.Problem / opportunityWhen reviewing documents or records together, users need a clear way to distinguish their status. For example, they may need to know whether a document is managed, whether it represents the latest revision, or whether an associated inspection is overdue.Status-based bounding box indicators would bring that context into the visual review, rather than requiring a separate status check for every item.Use Case Examples1. Managed versus unmanaged documentsUse different bounding box colors to distinguish documents identified as managed from those identified as unmanaged.Example: A user reviewing several documents could immediately identify wh
Its a need to have the schedules tab available for all users using InField. As a technician it is beneficial to see all upcoming check lists in a schedule view just as the lead technician is able to today. This gives a much better overview of all lists coming up in the next week.The only downside is that the technicians don’t need to set a check list to “Ready” status. Its proposed to give technicians a display version of schedules, without the possibility of setting lists to “Ready”.
Hey!We are looking into how to improve access control in our GitHub monorepo which deploys using Cognite Toolkit. We use a GitHub CODEOWNERS file to configure which teams approve changes to which code. Here are some options I see and the problems that I see with them:Option 1: Using different deployment credentials for different sets of Cognite Toolkit modulesThis initially seemed to me like the most natural option as we then use the access control system of the target platform, but we suspect that dealing with deployment ordering is then going to be too complicated for us to handle unless there is support for this approach from Cognite.Option 2: Checking that YAML source files do not go out of the bounds of their modulesWe then check that source files in a given module do not, for example, deploy into a CDF space managed by a different team, or reference environment variables containing secrets fetched from another team’s vault.My concern with this approach is that Cognite Toolkit’s v
Today there is no way to customize the way instances are ordered within hierarchy views. It would be beneficial with a method that could allow instances to be ordered to reflect the physical build of the monitoring system. Granted the information order can be delivered from the source into CDF, we can re-use the information to create a hierarchy view that more closely represents reality and what users may already be familiar with. However, in our case, we may have a sub-hierarchy (e.g. S1 data for Yggdrasil - Yggdrasil being the root of the hierarchy) that has one way of defining the order, while a different sub-hierarchy (e.g. SAS data for Yggdrasil) may not have exposed this information.
Welcome to the CDF Fundamentals Discussion!This discussion is dedicated to help learners of the Cognite Data Fusion Fundamentals learning path succeed. If you’re struggling with the exercises in this learning path, try the tips & tricks below or post a comment with the challenge you’re facing. You can also post your own tips and respond to fellow learners’ questions. Cognite Academy’s instructors are also here to help.
Hi Cognite team,>> On project `essc-sandbox-44` (cluster `westeurope-1`), we've configured a published AI Agent's data tools with **"Follow access scope defined manually"** set to a single instance space (`beveragemanufacturer`), but the tools scan the entire project's cross-space instance population instead of the scoped space. We verified via the SDK that the agent's "Found data of type" counts match the global totals exactly (e.g., 31,120 Process Alarms and 39,260 Downtime Events project-wide) while the scoped space holds only ~334 instances total — producing 6+ minute responses. It reproduces across three different tool types and is unaffected by reducing the data model, removing tools, or adding explicit prompt-level scoping.>> The attached document has full environment details, the verified evidence, reproduction steps, and our specific questions. Our main question: is this scope setting meant to filter DMS retrieval by space, and if so, is the cross-space scan a know
Already have an account? Login
No account yet? Create an account
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.
Check the
documentation
Ask the
Community
Take a look
at
Academy
Cognite
Status
Page
Contact
Cognite Support
