MLflow MCP
An MCP server providing end-to-end coverage for MLflow 3, including experiments, runs, registered models, model versions, logged models, traces, and assessments. It supports token-efficient extraction, run comparisons, failure mode clustering, and aggregation tools for AI debugging, available via stdio and Streamable HTTP transports.
Package and access
Local package with source-attributed installation metadata. Koskamo does not execute packages during review.
Checks and signals
Each line is separate evidence, not a trust score.
- Popularity 1 GitHub star
What you can do with it
- Find the run with the highest validation accuracy in an experiment
- Group trace failures by exception type and surface common errors
- Compare top runs by validation loss and show differing hyperparameters
- Get the latest version of a registered model with its training metrics
Source and maintenance
- Last repository push
- Latest release
- v1.12.7
- Maintenance checked
Packages and installation
Source-provided package data. Koskamo has not executed or live-probed this package.
@us-all/mlflow-mcp 1.12.7
{
"version": "1.12.7",
"transport": "stdio",
"identifier": "@us-all/mlflow-mcp",
"registryType": "npm",
"environmentVariables": [
{
"name": "MLFLOW_TRACKING_URI",
"isRequired": true,
"description": "MLflow tracking server URI (e.g. http://localhost:5050, https://mlflow.example.com)."
},
{
"name": "MLFLOW_TRACKING_TOKEN",
"description": "Bearer token for authenticated MLflow servers (e.g. Databricks)."
},
{
"name": "MLFLOW_TRACKING_USERNAME",
"description": "Basic-auth username (alternative to token)."
},
{
"name": "MLFLOW_TRACKING_PASSWORD",
"description": "Basic-auth password (alternative to token)."
},
{
"name": "MLFLOW_EXPERIMENT_ID",
"description": "Default experiment ID for run-creation tools."
},
{
"name": "MLFLOW_TOOLS",
"description": "Comma-separated category allowlist. Default: all categories enabled."
},
{
"name": "MLFLOW_DISABLE",
"description": "Comma-separated category disablelist."
},
{
"name": "MLFLOW_ALLOW_WRITE",
"default": "false",
"description": "Set to 'true' to enable write/destructive tools. Default read-only."
}
]
}Tools
Reported by the attributed source; not independently observed by Koskamo.
Maintenance history
Public check and source-refresh events. Editorial and ownership records remain private.
- · repository · passed · metadata changed · updated
- · backfill · recorded · Pre-expansion baseline; unobserved rich metadata remains unknown
Connect from your client
Each snippet configures the same source-provided package; replace placeholders before use.
Cursor
{
"mcpServers": {
"mlflow-mcp": {
"command": "npx",
"args": [
"-y",
"@us-all/mlflow-mcp@1.12.7"
],
"env": {
"MLFLOW_TRACKING_URI": "<YOUR_MLFLOW_TRACKING_URI>",
"MLFLOW_TRACKING_TOKEN": "<YOUR_MLFLOW_TRACKING_TOKEN>",
"MLFLOW_TRACKING_USERNAME": "<YOUR_MLFLOW_TRACKING_USERNAME>",
"MLFLOW_TRACKING_PASSWORD": "<YOUR_MLFLOW_TRACKING_PASSWORD>",
"MLFLOW_EXPERIMENT_ID": "<YOUR_MLFLOW_EXPERIMENT_ID>",
"MLFLOW_TOOLS": "<YOUR_MLFLOW_TOOLS>",
"MLFLOW_DISABLE": "<YOUR_MLFLOW_DISABLE>",
"MLFLOW_ALLOW_WRITE": "<YOUR_MLFLOW_ALLOW_WRITE>"
}
}
}
}Cline
{
"mcpServers": {
"mlflow-mcp": {
"command": "npx",
"args": [
"-y",
"@us-all/mlflow-mcp@1.12.7"
],
"env": {
"MLFLOW_TRACKING_URI": "<YOUR_MLFLOW_TRACKING_URI>",
"MLFLOW_TRACKING_TOKEN": "<YOUR_MLFLOW_TRACKING_TOKEN>",
"MLFLOW_TRACKING_USERNAME": "<YOUR_MLFLOW_TRACKING_USERNAME>",
"MLFLOW_TRACKING_PASSWORD": "<YOUR_MLFLOW_TRACKING_PASSWORD>",
"MLFLOW_EXPERIMENT_ID": "<YOUR_MLFLOW_EXPERIMENT_ID>",
"MLFLOW_TOOLS": "<YOUR_MLFLOW_TOOLS>",
"MLFLOW_DISABLE": "<YOUR_MLFLOW_DISABLE>",
"MLFLOW_ALLOW_WRITE": "<YOUR_MLFLOW_ALLOW_WRITE>"
}
}
}
}VS Code
{
"servers": {
"mlflow-mcp": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@us-all/mlflow-mcp@1.12.7"
],
"env": {
"MLFLOW_TRACKING_URI": "<YOUR_MLFLOW_TRACKING_URI>",
"MLFLOW_TRACKING_TOKEN": "<YOUR_MLFLOW_TRACKING_TOKEN>",
"MLFLOW_TRACKING_USERNAME": "<YOUR_MLFLOW_TRACKING_USERNAME>",
"MLFLOW_TRACKING_PASSWORD": "<YOUR_MLFLOW_TRACKING_PASSWORD>",
"MLFLOW_EXPERIMENT_ID": "<YOUR_MLFLOW_EXPERIMENT_ID>",
"MLFLOW_TOOLS": "<YOUR_MLFLOW_TOOLS>",
"MLFLOW_DISABLE": "<YOUR_MLFLOW_DISABLE>",
"MLFLOW_ALLOW_WRITE": "<YOUR_MLFLOW_ALLOW_WRITE>"
}
}
}
}Claude Desktop
{
"mcpServers": {
"mlflow-mcp": {
"command": "npx",
"args": [
"-y",
"@us-all/mlflow-mcp@1.12.7"
],
"env": {
"MLFLOW_TRACKING_URI": "<YOUR_MLFLOW_TRACKING_URI>",
"MLFLOW_TRACKING_TOKEN": "<YOUR_MLFLOW_TRACKING_TOKEN>",
"MLFLOW_TRACKING_USERNAME": "<YOUR_MLFLOW_TRACKING_USERNAME>",
"MLFLOW_TRACKING_PASSWORD": "<YOUR_MLFLOW_TRACKING_PASSWORD>",
"MLFLOW_EXPERIMENT_ID": "<YOUR_MLFLOW_EXPERIMENT_ID>",
"MLFLOW_TOOLS": "<YOUR_MLFLOW_TOOLS>",
"MLFLOW_DISABLE": "<YOUR_MLFLOW_DISABLE>",
"MLFLOW_ALLOW_WRITE": "<YOUR_MLFLOW_ALLOW_WRITE>"
}
}
}
}Let your agent find its tools
Find servers like MLflow MCP from your agent. Add Koskamo once to search the index for endpoints, access requirements, and client setup. Public discovery needs no key.