--- title: Introducing Kubeflow MCP: An Agent Interface for Cloud... description: Kubeflow has released an open MCP (Model Context Protocol) server that lets AI agents directly operate Kubeflow infrastructure—submitting training jobs,... canonical: https://daily.dev/posts/introducing-kubeflow-mcp-an-agent-interface-for-cloud-native-ai-at-scale-z8dfx6nql twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Introducing Kubeflow MCP: An Agent Interface for Cloud Native AI at Scale | daily.dev og:description: Kubeflow has released an open MCP (Model Context Protocol) server that lets AI agents directly operate Kubeflow infrastructure—submitting training jobs,... og:url: https://daily.dev/posts/introducing-kubeflow-mcp-an-agent-interface-for-cloud-native-ai-at-scale-z8dfx6nql og:image: https://api.daily.dev/og/posts/z8dfX6nqL.png og:image:alt: Introducing Kubeflow MCP: An Agent Interface for Cloud Native AI at Scale og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Introducing Kubeflow MCP: An Agent Interface for Cloud Native AI at Scale **[Kubeflow](https://daily.dev/sources/kubeflow)** · 8 min read · 8 upvotes · 0 comments ## Summary Kubeflow has released an open MCP (Model Context Protocol) server that lets AI agents directly operate Kubeflow infrastructure—submitting training jobs, checking GPU resources, and streaming logs—without a human copy-pasting commands and errors between an agent and a cluster. The server exposes 23 tools across planning, discovery, training, monitoring, lifecycle, and platform phases, with two-phase confirmation gating any job-creating or destructive action. It ships with four personas (readonly, data-scientist, ml-engineer, platform-admin), three token-efficient tool-loading modes (full, progressive, semantic), platform auto-detection for OpenShift/EKS/GKE/Kind, rate limiting, circuit breakers, and OpenTelemetry tracing. Currently only the Kubeflow Trainer client is implemented (requires Trainer v2.2+, Kubernetes 1.27+, Python 3.10–3.12), with Optimizer/Katib, Model Registry, Pipelines, Spark Operator, and Feast planned as future MCP clients. Setup instructions cover installing via pip and connecting from Cursor IDE or Claude Code. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is the Kubeflow MCP server and what does it let AI agents do? It is an open-source MCP (Model Context Protocol) server that exposes Kubeflow operations as tools AI agents can call directly, such as submitting training jobs, checking cluster GPU resources, and streaming training logs. It ships with 23 tools across planning, discovery, training, monitoring, lifecycle, and platform-admin phases, and currently supports the Kubeflow Trainer client, with Optimizer, Model Registry, Pipelines, Spark Operator, and Feast planned. _daily.dev helps developers evaluating agent-native ML infrastructure tools track releases like this one._ ### What are the prerequisites to install and run kubeflow-mcp? You need Python 3.10 to 3.12, Kubernetes 1.27 or later, and a cluster with Kubeflow Trainer v2.2 or later installed. Install via pip install kubeflow-mcp, then run kubeflow-mcp serve with flags for persona (readonly, data-scientist, ml-engineer, platform-admin), mode (full, progressive, semantic), and transport (stdio, http, sse). _developers setting up MCP-based ML tooling can follow release notes like this via daily.dev._ ### How does two-phase confirmation prevent AI agents from accidentally destroying cluster resources? Every job-creating or destructive operation, such as training submission, job deletion, or runtime create, patch, and delete, previews its full effect first and only proceeds after explicit user approval. This ensures an agent can never accidentally create or destroy a resource without the user seeing the complete spec beforehand, keeping a human in the loop for high-stakes actions. _engineers weighing safety guardrails for agent-driven infrastructure changes can find design patterns like this on daily.dev._ ## Similar posts on daily.dev - [Building effective AI agents with Model Context Protocol \(MCP\)](https://daily.dev/posts/building-effective-ai-agents-with-model-context-protocol-mcp--1k6pmnpvj) · Red Hat Developer · 1 upvotes · 0 comments - [Red Hat Readies an MCP Server to Help LLMs Manage Kubernetes](https://daily.dev/posts/red-hat-readies-an-mcp-server-to-help-llms-manage-kubernetes-pn6o59xt9) · Container Journal · 0 upvotes · 0 comments - [Build AI agents faster with GCS \(Google Cloud Storage\) MCP server](https://daily.dev/posts/build-ai-agents-faster-with-gcs-google-cloud-storage-mcp-server-ysjglamng) · Google Cloud · 1 upvotes · 0 comments - [Playing with Snowflake-managed MCP](https://daily.dev/posts/playing-with-snowflake-managed-mcp-tn6no5rel) · Snowflake Community · 1 upvotes · 0 comments --- Tags: [#kubernetes](https://daily.dev/tags/kubernetes), [#ai-agents](https://daily.dev/tags/ai-agents), [#mcp](https://daily.dev/tags/mcp), [#mlops](https://daily.dev/tags/mlops) [View this post on daily.dev](https://daily.dev/posts/introducing-kubeflow-mcp-an-agent-interface-for-cloud-native-ai-at-scale-z8dfx6nql) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Introducing Kubeflow MCP: An Agent Interface for Cloud Native AI at Scale","url":"https://daily.dev/posts/introducing-kubeflow-mcp-an-agent-interface-for-cloud-native-ai-at-scale-z8dfx6nql","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/introducing-kubeflow-mcp-an-agent-interface-for-cloud-native-ai-at-scale-z8dfx6nql"},"datePublished":"2026-08-12T15:54:34.521Z","dateModified":"2026-09-14T06:17:08.716Z","description":"Kubeflow has released an open MCP (Model Context Protocol) server that lets AI agents directly operate Kubeflow infrastructure—submitting training jobs,...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/e9777282069100831dfb326940a5b097?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/e9777282069100831dfb326940a5b097?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Kubeflow","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Kubeflow","logo":"https://media.daily.dev/image/upload/s--kSOGBli9--/f_auto/v1743318307/logos/kubeflow","url":"https://daily.dev/sources/kubeflow"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/introducing-kubeflow-mcp-an-agent-interface-for-cloud-native-ai-at-scale-z8dfx6nql","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":8},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"kubernetes,ai-agents,mcp,mlops","timeRequired":"PT8M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Kubeflow","item":"https://daily.dev/sources/kubeflow"},{"@type":"ListItem","position":3,"name":"Introducing Kubeflow MCP: An Agent Interface for Cloud Native AI at Scale"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/introducing-kubeflow-mcp-an-agent-interface-for-cloud-native-ai-at-scale-z8dfx6nql#faq","mainEntity":[{"@type":"Question","name":"What is the Kubeflow MCP server and what does it let AI agents do?","acceptedAnswer":{"@type":"Answer","text":"It is an open-source MCP (Model Context Protocol) server that exposes Kubeflow operations as tools AI agents can call directly, such as submitting training jobs, checking cluster GPU resources, and streaming training logs. It ships with 23 tools across planning, discovery, training, monitoring, lifecycle, and platform-admin phases, and currently supports the Kubeflow Trainer client, with Optimizer, Model Registry, Pipelines, Spark Operator, and Feast planned. daily.dev helps developers evaluating agent-native ML infrastructure tools track releases like this one."}},{"@type":"Question","name":"What are the prerequisites to install and run kubeflow-mcp?","acceptedAnswer":{"@type":"Answer","text":"You need Python 3.10 to 3.12, Kubernetes 1.27 or later, and a cluster with Kubeflow Trainer v2.2 or later installed. Install via pip install kubeflow-mcp, then run kubeflow-mcp serve with flags for persona (readonly, data-scientist, ml-engineer, platform-admin), mode (full, progressive, semantic), and transport (stdio, http, sse). developers setting up MCP-based ML tooling can follow release notes like this via daily.dev."}},{"@type":"Question","name":"How does two-phase confirmation prevent AI agents from accidentally destroying cluster resources?","acceptedAnswer":{"@type":"Answer","text":"Every job-creating or destructive operation, such as training submission, job deletion, or runtime create, patch, and delete, previews its full effect first and only proceeds after explicit user approval. This ensures an agent can never accidentally create or destroy a resource without the user seeing the complete spec beforehand, keeping a human in the loop for high-stakes actions. engineers weighing safety guardrails for agent-driven infrastructure changes can find design patterns like this on daily.dev."}}]} ```