Integration

Dataiku DSS integration

Dataiku DSS is a data science and machine learning platform; each customer works on its own DSS instance. An agent can list projects, datasets, recipes, scenarios and managed folders; read dataset schemas, metadata, partitions and last metric values and recipe code; start and abort jobs that build datasets on the customer's DSS and read their status and logs; run and abort scenarios and follow their runs, step outcomes and logs; update project metadata; and list, download, upload and delete files in managed folders. Not exposed: dataset rows (an unbounded stream), connection settings and plugin parameters of datasets, recipes and folders, project and scenario variables, full scenario settings, activating or deactivating scenarios (its route also writes a scenario's description, tags and checklists), permissions, creating or deleting projects, datasets and recipes, machine learning, deployer and administration routes. instance_url is the address of your DSS server with its scheme and port (for example https://dss.mycompany.com:11200). Authentication is a DSS API key sent as a Bearer token: a personal key (Profile & Settings > API keys) acts as its user, a project key (Project Home > Security > API Keys) or a global key (Administration > Security) has only the privileges set on it.

Connect Dataiku DSS How integrations work

Ask it in chat

Say what should happen.
The workflow builds itself.

When this happens, do that in Dataiku DSS.Built from the 30 Dataiku DSS tools below, on the connection you add once.

What you can do with Dataiku DSS

Each of these is one operation on one endpoint, available both as a workflow step and as a tool an AI agent can call on its own. Nothing bundles several operations into one action, so what a run did stays readable afterwards.

Most endpoints run on the free plan. A few, mainly the ones that publish on your behalf, need a paid plan.

Content, 4 operations

delete_folder_fileDelete folder fileDELETEDeletes one file of a managed folder. Requires WRITE_DATA on the folder.
download_folder_fileDownload folder fileGETDownloads one file of a managed folder and returns it as a stored file. Requires READ_DATA on the folder.
list_folder_contentsList folder contentsGETLists the files of a managed folder with each file's path, size and last modification time. Requires READ_DATA on the folder.
upload_folder_fileUpload folder filePOSTUploads a file to a managed folder at the given path, overwriting a file already there. Requires WRITE_DATA on the folder.

Metadata, 4 operations

get_dataset_metadataGet dataset metadataGETReturns a dataset's metadata, its display label, description, tags and custom metadata. Requires READ_METADATA on the dataset.
get_project_metadataGet project metadataGETReturns a project's metadata: its display name (label), short and long descriptions, tags, custom metadata, checklists and custom fields. Requires READ_CONF on the project.
get_recipe_metadataGet recipe metadataGETReturns a recipe's metadata, its display label, description, tags and custom metadata. Requires READ_METADATA on the project.
update_project_metadataUpdate project metadataPUTReplaces a project's whole metadata object. Send back every field get_project_metadata returned (label, description, shortDesc, tags, custom, checklists, customFields) with your changes; a field left out is cleared. Requires ADMIN.

Job, 3 operations

get_job_statusGet job statusGETReturns the status of a job, a sequence of activities (one recipe run on a set of partitions each). Its state is NOT_STARTED or RUNNING while in flight, then DONE, FAILED or ABORTED. Requires MONITOR_JOBS.
list_jobsList jobsGETLists the latest jobs of a project with their id, state, what they built and when. Requires READ_CONF on the project.
start_jobStart jobPOSTStarts a job that builds datasets or other outputs, running their recipes on the customer's DSS infrastructure. Returns the job id at once; follow it with get_job_status. Requires RUN_JOBS on the project.

Abort, 2 operations

abort_jobAbort jobPOSTAsks a running job to abort. The job may take some time to actually stop, so read get_job_status until its state is final. Requires RUN_JOBS on the project.
abort_scenarioAbort scenarioPOSTAborts the current run of a scenario; it does nothing when the scenario is not running. Requires RUN_JOBS on the project.

Log, 2 operations

get_job_logGet job logGETReturns the log of a job as plain text, or the log of one of its activities. Useful to find why a job FAILED. Requires READ_CONF on the project.
get_scenario_run_logGet scenario run logGETReturns the log of a scenario run as plain text, or the log of one of its steps. Useful to find why a run FAILED. Requires READ_CONF on the project.

Managedfolder, 2 operations

get_managed_folderGet managed folderGETReturns a managed folder's id, name, storage path and tags. Requires READ_CONF on the project.
list_managed_foldersList managed foldersGETLists the managed folders of a project (folders of files the Flow reads and writes) with their id, name, path and tags. Requires READ_CONF on the project.

Recipe, 2 operations

get_recipeGet recipeGETReturns a recipe's definition (type, inputs and outputs) and its payload, the code or the visual settings it runs. Requires READ_CONF on the project.
list_recipesList recipesGETLists the recipes of a project (the steps of its Flow) with their name, type and input and output datasets or folders. Requires READ_CONF on the project.

Scenario, 2 operations

get_scenario_runGet scenario runGETReturns the details of one scenario run, its start, end and outcome, and the outcome of each step it launched, including the ids of the jobs those steps executed. Requires RUN_JOBS on the project.
list_scenariosList scenariosGETLists the scenarios of a project with their id, name, whether each is running now and whether it is active (responding to its own triggers). Requires MONITOR_JOBS on the project.

Dataset, 1 operation

list_datasetsList datasetsGETLists the datasets of a project with their name, type, format, tags and connection (name and path only). Requires READ_CONF on the project.

Get last run, 1 operation

list_scenario_runsList scenario runsGETLists the last runs of a scenario, newest first, with their start, end and outcome; a run still in progress has no result yet. Requires RUN_JOBS on the project.

Get run for trigger, 1 operation

get_scenario_run_for_triggerGet scenario run for triggerGETReturns the scenario run a trigger fire started, from the trigger.id and runId run_scenario returned. scenarioRun is absent until the run has started; read it again a few seconds later. Requires RUN_JOBS.

Last, 1 operation

get_dataset_last_metricsGet dataset last metricsGETReturns the last computed value of every metric on a dataset or one of its partitions (such as its size), with when each was computed. Requires READ_METADATA.

Light, 1 operation

get_scenario_statusGet scenario statusGETReturns whether a scenario is running now and whether it is active (responding to its own triggers). Requires MONITOR_JOBS on the project.

Partition, 1 operation

list_dataset_partitionsList dataset partitionsGETLists the partition identifiers of a dataset, such as 2015-01-01; a dataset that is not partitioned returns the single value NP. Requires READ_DATA on the dataset.

Project, 1 operation

list_projectsList projectsGETLists the keys of the DSS projects on which the API key holds the READ_CONF privilege, optionally only those carrying one of the given tags. Read a project's display name with get_project_metadata.

Run, 1 operation

run_scenarioRun scenarioPOSTStarts a scenario on the customer's DSS infrastructure by firing its manual trigger. Returns at once with the trigger fire; pass trigger.id and runId to get_scenario_run_for_trigger to follow the run. Requires RUN_JOBS.

Schema, 1 operation

get_dataset_schemaGet dataset schemaGETReturns the schema of a dataset, the list of its columns with their types. Requires READ_SCHEMA on the dataset.

Use Dataiku DSS from Claude Code, Cursor or Codex

Connect Dataiku DSS in LiveContext, create an API key in Settings, then add the LiveContext MCP server to your client. Its Dataiku DSS operations become tools the client can call, next to every other integration you connected.

Claude Code

claude mcp add --transport http livecontext https://livecontext.ai/mcp --header "X-API-Key: YOUR_API_KEY"

Codex (~/.codex/config.toml)

[mcp_servers.livecontext]
url = "https://livecontext.ai/mcp"

[mcp_servers.livecontext.http_headers]
"X-API-Key" = "YOUR_API_KEY"

Cursor, Claude Desktop and other JSON configs

{
  "mcpServers": {
    "livecontext": {
      "type": "http",
      "url": "https://livecontext.ai/mcp",
      "headers": {
        "X-API-Key": "YOUR_API_KEY"
      }
    }
  }
}

Set up the MCP server step by step

How to connect Dataiku DSS

Create your workspace
Sign up free. Nothing to install.
Connect Dataiku DSS once
Add your Dataiku DSS connection in Settings. Every tool from this integration reuses it, and secrets are encrypted at rest and visible only to you.
Describe the job in chat
The workflow builds itself in front of you and picks the endpoints it needs. Drag the steps yourself on the canvas whenever you would rather be explicit.