• PATHING
  • VISUALS
  • DATA
  • HUMAN

How to locate model artifacts with your AI

Connect MLflow to your agent to pull specific files or graphs from your experiment runs. Once connected, you can ask for exact paths or data blobs without digging through the MLflow UI.

Ask AI about this page

Short answer

How can I find specific files or graphs in MLflow?

Your agent uses the list_artifacts tool to scan experiment runs and identify the exact location of files, blobs, or plots. You'll get direct paths and data locations for your models and metrics.

Where your data is found

What you can do with your agent

The agent handles the heavy lifting of navigating your MLflow experiment structure.

PATHING

Find exact file paths

The agent identifies the specific directory or blob location for any model component. You get the direct path needed for loading or moving files.

VISUALS

Locate visualization graphs

Your agent finds the saved plots and metric visualizations within a run. This helps you quickly grab the right image files for reports.

DATA

Identify data blobs

The agent scans for specific data files or serialized objects stored in the run. You'll know exactly where your training data or weights are sitting.

HUMAN

Manual review needed

You still need to manually inspect the actual content of the files or plots. The agent finds the files, but you decide if the data looks right.

The workflow

What your AI does when the request arrives.

The agent moves from identifying the run to pinpointing the specific file you need.

  1. Identify the run

    The agent looks up the specific experiment and run ID you mentioned to establish the correct context.

    list_artifacts

  2. Scan for files

    The agent crawls the directory structure of that run to find all available files and folders.

    list_artifacts

  3. Filter for targets

    The agent narrows down the list to find the specific model weights, config files, or plots you asked for.

    list_artifacts

  4. Provide paths

    The agent returns the precise artifact paths so you can use them in your code or client.

    list_artifacts

Try it

Copy these to start.

Use these prompts to begin interacting with your MLflow data.

Starting points

Replace the bracketed text with your actual experiment names or run IDs.

MLflow Connector

You're all set. Choose your MCP client and follow the setup instructions.

Connector linkhttps://edge.vinkius.com/vk_preview_O0ZBpwFtJKVOQRXy98KSjGZvQNA27XEpDUDcFNyB/mcp

Claude Desktop

Follow the steps below to connect in seconds.

  1. 1In Claude Desktop, open Settings → Connectors.
  2. 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
  3. 3Click Add and start a new chat — MLflow capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "mlflow-ml-lifecycle-management-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_O0ZBpwFtJKVOQRXy98KSjGZvQNA27XEpDUDcFNyB/mcp"
    }
  }
}
Copy into chat04

  • List all the artifacts in the experiment run 'model_training_v2'.

  • Find the path to the model weights file in run 'abc-123-xyz'.

  • What visualization plots are available in the 'hyperparameter_tuning' experiment?

  • Show me the file structure for the artifacts in run 'run_id_9988'.

  • Claude
  • ChatGPT
  • Cursor
  • VS Code
  • Windsurf
  • Claude Code
  • JetBrains
  • Cline

Start here

Connect MLflow once, then ask.

One connection gives your agent access to your MLflow workspace. Your credentials stay encrypted and secure on our side.

Connect MLflow to your AI

FAQ

How this task behaves

  • 01

    Can the AI download the files for me?

    No. The agent finds the paths and locations of the files, but you must handle the actual downloading or loading in your own environment.

  • 02

    Can the AI delete artifacts from my MLflow server?

    No. The agent only has permission to list and find artifacts. It cannot modify, delete, or overwrite any data in your MLflow runs.

  • 03

    Does it work with any MLflow backend?

    Yes, as long as your MLflow instance is accessible through the connection you set up.

  • 04

    Can the AI create new artifacts?

    No. The agent is limited to reading and listing existing artifacts via the provided tools.

  • 05

    Can I find specific versions of a model?

    Yes, if those versions are stored as distinct artifacts within a specific run, the agent can locate them.

  • More questions about MLflow? See everything the MLflow Connector can do

Connect MLflow to Claude, Cursor, ChatGPT & more