News & Intel
LiveDaily AI-curated intelligence on Palantir Foundry, Ontology, AIP, Apollo, contracts, and community feedback. Updated automatically via GitHub Actions every day at 7 AM UTC.
Unknown error of type "invalidObjectTypeApiName"
This error just populated and the pipeline won’t allow me to deploy. I haven’t touched anything related to API name. The object API name was automatically generated when it was first built. I asked AIP assist in order to solve this issue. The way ahead from AIP assist is to locate API name from the dataset or transform in the pipeline and compare with the object’s API name. However, I couldn’t location the API name from the dataset or transform. Does anyone ran into this issue by any chance? 2 posts - 2 participants Read full topic
Stop regressing pytorch-gpu for pytorch
When I install PyTorch-gpu via conda in vs code, I do not mean to regress to vanilla pytorch whenever Foundry deems it appropriate. Fix it please. 1 post - 1 participant Read full topic
Feature Idea: A "Stack Watcher" agent for the enrollment
AIP and AI FDE have shown how powerful LLM agents can be when acting directly on the Palantir stack: building transforms, editing the ontology, debugging pipelines, and managing repos. It genuinely feels like having an extra forward-deployed engineer who never sleeps. If AI FDE is the builder, the Stack Watcher should be the foreman. A persistent, always-on observer that quietly monitors the entire enrollment and delivers a concise morning brief. Concrete version: I come back from a week of PTO, and my first hour is reconstructing what shipped, what broke, and what someone needs me to know. The Watcher collapses that to two minutes. What the daily brief should cover Narrative summary of changes. Not a raw log, but the actual story: “Team Alpha merged three branches optimizing the Supply Chain ontology; the new marking on Sensitive_Orders may need a second look.” Backed by concrete deltas (new datasets, ontology edits, permission grants, merged branches, deployed apps, r
Relabel Values in Object
I have a Chart XY in Workshop that I am sorting by Value Ascending to match the process order. An example of the values from the original column in the Object are Start, Middle, and End. I created a new column to correctly order them so that when you are viewing the chart you are seeing them in the process order. The updated values are “1_start”, 2_middle", and “3_end”. This solves the issue of sorting the values but I would like to be able to relabel the values so that they look cleaner. I know that there is Conditional Formatting within objects but I think that only allows you to update the color or the alignment, not the way that the value is presented. Any suggestions on how to relabel values? Thanks! 1 post - 1 participant Read full topic
Hide X-axis label name in Workshop chart
Hi Team, I am working on a Workshop application where I have created a bar chart with Work Center on the X-axis (Assembly, Avionics Lab, Machining, etc.). I want to hide the X-axis label name (for example, “Work Center”) but still display the category values on the axis. Could anyone please guide me on how to remove or hide only the X-axis title/label? Thanks in advance! 2 posts - 2 participants Read full topic
Running PySpark Tests in Code Workspaces
What’s the preferred pattern to run Spark DataFrame based tests using Code-Workspaces please? We have a few code respositories that run tests using the pattern described in the Docs DataFrame Unit Test Example , e.g. def test_dataframe(spark_session): df = spark_session.createDataFrame([['a', 1], ['b', 2]], ['letter', 'number']) assert df.schema.names == ['letter', 'number'] But this doesn’t work in Code-Workspaces, the tests fail because there is no spark_session available. I can find the existing fixture for this which is at site-packages/foundry_pyls/fixtures.py (code not posted because of copyright notice) and add that into my own conftest.py file as a PyTest Fixture, but it doesn’t work. It additionally requires installing the ‘Pyspark’ Python package, but: That doesn’t seem to work, even with Pyspark installed it can’t find Spark Adding Pyspark directly like this seems to conflict with the Foundry Transforms imports, causing other issues. How c
Displaying function-backed aggregated values alongside object properties in an Object Table
I have an Object Table in Workshop displaying objects from a dataset. I want to show: A grouping property (e.g. a product name or account name) Aggregated SUM of numeric metrics per group (e.g. total revenue, total quantity) I have a TypeScript function that computes the top N groups by an aggregated metric and returns the summed values. However: Object Table only accepts ObjectSet or Osdk.Instance — it can only display properties that exist on the object type, not computed/aggregated values. What I need: A table where each row represents a grouped value, showing: The grouping property value Function-computed aggregated metrics Is there a supported way to display function-backed aggregated columns in a table widget, or does this require a pre-aggregated object type via a pipeline? 1 post - 1 participant Read full topic
Deduplicated View for Linked Object Properties in Workshop Module
Hi, Below the details for this feature request. Problem: In the Workshop module’s object table, adding a linked object property can result in hundreds of duplicated values depending on the number of related objects and their content. This makes it difficult to identify distinct values and reduces usability. The current workaround requires creating a custom function to deduplicate results. However, this must be implemented separately for each object and each link, which becomes inefficient and time-consuming in environments with many relationships. Proposed Solution: Introduce a built-in option to display distinct (deduplicated) values for linked object properties directly within the object table. This could be implemented as a toggle or setting (e.g., “Show unique values only”) applied at the property level. Use Cases: Viewing a clean list of unique linked values without duplicates Simplifying analysis of relationships between objects Reducing the ne
GeoHash -> GeoPoint in pipeline builder?
What is the most efficient way to go from a geohash → geopoint in pipeline builder? I see a geopoint → geohash expression, but not the other way around 2 posts - 2 participants Read full topic
Bug Report: Media Sets not registering on main
I’m getting a build failure caused by: “MediaSet:CannotReadMediaSetBranch”. According to Data Lineage, the media set does not exist on main (see screenshot below). In a simplified pipeline pasted below, Media Set XLSX Output can build, but Dataset XLSX Output fails to access a media set item (CannotReadMediaSetBranch) after converting media reference to a logical type: Media Reference. For context, I need to develop a drop zone where users can upload the latest Excel extract from a business system. I tried to use a media set backed property in an ontology action type, and this error is blocking me from using the Excel files from the materialization for downstream processing. Please advise both about steps to resolve this error and alternative workflows. Potentially related: Foundry branching: medias sets on branches 2 posts - 2 participants Read full topic
FR: Reset / Discard user edits and fall back to backing dataset
Hi, we are very frequently stumbling across a missing functionality, which is the reset/discard user edits on object types. Imagine, you have an object type that is backed by a backing dataset. It has a property that frequently updates via pipeline, but you want to allow users to edits that value. The underlying dataset is still frequently updating the value, but since it has an ontology edit, only that edit is valid - so far so good. Now imagine the case, the user wants to reset that edit (discard) and let again the value be defined by the frequently-updating backing dataset value. As per my understanding, there is actually no direct way of achieving such a discard. In situations where we can not prevent the discard case, we are implement a very effortfull workaround This is: we store edits which potentially require reset in separate object types additionally. When the user applies an edit, we edit the main object type and create the secondary „edited property" object. With
Ontology as Code does not support using virtual tables when mapping backing datasource in Marketplace
I am attempting to install an updated version of my ontology as code marketplace application. I am attempting to switch to virtual tables that were created via a Snowflake connection. But the folder containing the datasets shows up blank which I take to mean they are unsupported. I am able to back the ontology objects with the same virtual tables from Ontology Manager if I first back them with notional datasets to get through the install process. Is this a known issue? When will support be added? 1 post - 1 participant Read full topic
Workshop Custom Typography Feature - Missing Documentation
I’m trying to use the “Custom typeface” feature in Workshop settings (under Typography section), but I can’t find any documentation on how to actually install and use custom fonts. What I see: Workshop Settings → Typography → Custom typeface dropdown Field asking to “Choose an installed typeface by typing its name” System fonts are available (Arial, Helvetica, etc.) But no instructions on how to add custom fonts What I need to know: How do I upload/install custom font files (.otf or .ttf)? Where in Control Panel or Foundry do I configure this? How do I make custom fonts available organization-wide for Workshop? What I’ve tried: Searched official Foundry documentation - only found custom font docs for Slate, not Workshop Checked Control Panel settings - no obvious font upload section I have admin access but still can’t figure it out Has anyone successfully used custom fonts in Workshop? If so, please s
Connecting Agents to Decisions
The Palantir Ontology Palantir’s software powers real-time, human-agent decision-making in many of the most critical commercial and government contexts around the world. From disaster response to nuclear energy production , our customers depend on Palantir AIP to safely, securely, and effectively leverage AI in their enterprises — and drive operational transformation. While many factors contribute to achieving and scaling operational impact, including our AIP AgentCamps — where customers are hands-on-keyboards and achieving outcomes with AI in a matter of hours — the key differentiator is a software architecture which revolves around the Palantir Ontology. The Ontology is a system designed to represent the decisions in an enterprise, not simply the data. The prime directive of every organization in the world is to execute the best possible decisions, often in real-time, while contending with internal and external conditions that are constantly in flux. Traditional data archit
AI FDE Error: Failed to load Workshop module due to missing Action Parameter RID
Context: I am a software designer who has been leveraging Claude Code extensively over the past year. I recently transitioned into Foundry and have been utilizing the AI FDE tool. Compared to manual configuration within Foundry, the experience has been excellent; I’ve successfully managed CRUD operations for object types, actions, and general Ontology metadata without issue. The Problem: I am now attempting to use AI FDE to iterate on our UI Modules (Workshop) . While I can reference the UI Module correctly, AI FDE fails during the “loading” phase with the following error: [Error] Failed to load Workshop module: No ID found for action parameter RID: ri.actions.main.parameter.14d4d52e-315f-4976-a7a7-92e261affbc3 (at config path: eb7403c7-16e7-4cd6-9c7e-2dab9b5c4526.actionConfigs.0.actionConfig) Troubleshooting Steps Taken: Baseline Validation: I created a fresh “sandbox” Workshop module to verify if AI FDE supports UI editing. This worked as expected—I wa
Additional Metrics in Workshop Granular Metrics
Would love to see more metrics being logged with Workshop > Metrics > Enable granular metrics. For instance, time and date, username, user department/branch, etc. At present only mid-level anonymized metrics are gathered (Actions vs. Submission Count, and Views (Pages, Overlays, Tabs) vs View Count, and the time frames are hard coded at 7, 30, 90 days. Customers are often interested in who used their app, what department they were from, when the app was used, and they may want to compare this month with last month, all things the present tool doesn’t support. A workaround can be implemented with hidden slate iframe that Workshop displays, and the Slate then runs a create log object action on page load, but if these features were built into Workshop Granular Metrics it would reduce complexity and increase maintainability (over the workaround with Slate iframes in Workshop to log). 1 post - 1 participant Read full topic
Question about object indexing, preserving the order / sorting of backing datasets
I tried sorting in pipeline builder a particular dataset, but when the object gets indexed the object’s order is completely different than the backing dataset’s. I initially thought this is a random shuffle but when I present the object in a table in workshop, there is a consistent order (unsorted) whenever I refresh the page. This brings up the question - in Palantir’s indexing process in an object set, how is the order determined? Is it completely random? Is there any way to preserve sort in pipeline builder to translate to the object level? (I feel like this defeats the purpose of sorting in pipeline builder)… Thank you! 1 post - 1 participant Read full topic
Expose Resource Lineage as a Public API
Foundry’s Workflow Lineage UI shows the full dependency graph between resources (datasets, object types, transforms, applications, monitors). This data is not accessible programmatically. Please expose it as a V2 REST API endpoint and MCP tool . Today this data is only available through the Workflow Lineage UI and AI FDE’s internal tooling. Functions, AIP Agents, and OSDK apps cannot access it. What We Tested From a TypeScript v2 Function, we tried 30+ API paths including Compass sub-paths, decorations, V2 endpoints, and all @osdk/foundry.* SDK packages. None expose lineage data. Meanwhile, AI FDE’s internal get_resource_lineage tool returns exactly what we need — upstream/downstream RIDs with names, paths, and connection types. The Ask REST API — GET /api/v2/resources/{rid}/lineage for Functions, OSDK apps, and scripts MCP Tool — get_resource_lineage for external AI agents (Claude Desktop, etc.) AIP Agent Platform Tool — Same capabi