Palantir MCP [Beta]

Palantir MCP is an implementation of the Model Context Protocol ↗. Palantir MCP enables AI IDEs and AI agents to autonomously design, build, edit, and review end-to-end applications within the Palantir platform, covering everything from data integration to ontology configuration and application development. In addition, you can use Palantir MCP to allow external AI systems to query documentation, metadata, and data, as well as perform high-level tasks on the platform. Developers can use Palantir MCP to automate auxiliary tasks while they stay focused on the system they are building.

Getting started

To get started with the Palantir MCP, first follow the Palantir MCP installation documentation.

Next, try one of the prompts below.

Ask for coding help

If you need coding help, the Palantir MCP can help by searching the Palantir documentation or getting context about your current repository.

In a TypeScript OSDK repository:

Copied!
1 What ontology objects do I have in my OSDK?

In a Python transforms repository:

Copied!
1 Can you rewrite this Python transform to use lightweight?

Analyze a dataset

If you are working in a Python transforms repository, it is often necessary to understand the contents of a dataset. You can prompt the model to analyze datasets for you by providing a dataset RID.

Replace the in the prompt below with a dataset RID, or reference an existing transform.

Copied!
1 Can you please find an appropriate primary key in this dataset: <RID>?

Create an Ontology object with mock data

If you are working on a TypeScript OSDK application, it is common to reference data that may not be in the Ontology yet. Instead of breaking focus on your current task, you can ask the MCP to create mock data for you and continue working on your application.

Copied!
1 2 3 4 5 I'm working on this OSDK application and need to reference some user data that I don't have in my OSDK yet. Can you please create a new mock user dataset, create a new User Ontology object, and then create a pull request for me? The user dataset should have the columns: user_id, name, email

Available tools

Palantir is continually developing more tools, prompts, and resources to improve Palantir MCP. Currently, Palantir MCP provides the following tools:

Tool nameDescription
create_and_write_to_foundry_datasetCreates a dataset in Foundry with data provided from a CSV file, useful for uploading local data or creating mock data.
create_foundry_branchCreates a new Foundry branch with a specified display name and description.
create_or_update_foundry_object_typeCreates or updates object types in the ontology on a provided branch.
delete_foundry_object_typeDeletes an object type from the ontology on a provided branch.
generate_new_ontology_sdk_versionAdds ontology entities modified by a given foundry proposal to the ontology SDK and generates a new version of that SDK.
get_foundry_dataset_schemaReturns the definition of a Palantir dataset for a given dataset RID or dataset path, including schema information.
get_foundry_ontology_ridFetches the Ontology RID for a given Foundry OSDK application or local repository.
get_ontology_sdk_contextFetches curated documentation about the Ontology SDK for the user's OSDK repository.
get_or_create_foundry_proposalCreates a new Foundry proposal, which is the final step of a Foundry branch.
get_python_transforms_documentationFetches curated documentation about Python transforms and the transforms.api SDK.
list_resources_in_foundry_folderReturns the first 20 resources available in a given folder or project RID, with pagination support.
run_sql_query_on_foundry_datasetReturns the result of running a SQL query on a Palantir dataset on an optional branch.
search_foundry_documentationSearches Palantir Foundry platform documentation indices for relevant context snippets for a given query.
view_foundry_branchAllows viewing an existing Foundry branch, including branched resources and ontology modifications.
view_foundry_object_typeAllows viewing an existing object type in the ontology.