# KubeDrive (mcpb · kubedrive-0.6.6.mcpb)

Your Kubernetes clusters for AI agents through the KubeDrive desktop app, without a kubeconfig.

- Trust score: 37/100 (low)
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-11-05

## Components

- mcpb · `kubedrive-0.6.6.mcpb`: 37/100 (this document), [markdown](https://verifymcp.io/servers/aleksskoz-kubedrive/kubedrive.md), [page](https://verifymcp.io/servers/aleksskoz-kubedrive/kubedrive)

## Channel facts

- Registry: `mcpb`
- Package: `https://github.com/aleksskoz/kubedrive/releases/download/v0.6.6/kubedrive-0.6.6.mcpb`
- Transport: `stdio`
- SHA-256: `0d96927e9d9821a86e143e1c0b2d283d903705ec08d1c16ebe0161f0cc2fb0c5`

## Trust breakdown

How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog)

Scored 2026-11-05.

- **Supply Chain Security**: 0/100
  - No malware scan is available for this kind of package: the supply-chain vendors we use do not cover it. This is a permanent gap in our coverage, not a finding about the package.
  - Known CVEs could not be checked: we found no dependency list to check in this bundle: either it runs a compiled binary, whose bundled manifests we do not read, or none of the package.json, requirements.txt, pyproject.toml or uv.lock files it ships for its runtime declares a dependency we can read (a requirements.txt that only points at a setup.py project, a local path, or a nested requirements file we cannot follow counts as none).
  - Install-script risk not yet assessed.
  - Dependency health could not be checked: we found no dependency list to check in this bundle: either it runs a compiled binary, whose bundled manifests we do not read, or none of the package.json, requirements.txt, pyproject.toml or uv.lock files it ships for its runtime declares a dependency we can read (a requirements.txt that only points at a setup.py project, a local path, or a nested requirements file we cannot follow counts as none).
- **Provenance & Transparency**: 32/100
  - Source repository is publicly reachable at the declared URL.
  - Provenance check failed: no build-provenance attestation is published.
  - License check failed: the license (SEE LICENSE IN https://github.com/aleksskoz/kubedrive/blob/main/LICENSE) isn't a recognized OSI-approved license.
  - Actively maintained (last published 1 days ago).
  - Disclosure check failed: no security disclosure policy was found in the source repository.
- **Schema Quality & AI Usability**: 66/100
  - AI-judged instruction clarity (good).
  - Context-footprint check failed: tool/resource definitions use about 3093 tokens (~118/item across 26 items; 26 tools + 0 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 20/100
  - Stability observed for 6 of 30 days with no destabilising changes; credit accrues until the full window elapses.
- **Tool Coverage**: 96/100
  - 100% of tools have a non-trivial description (not blank, and not just the tool's name).
  - 88% of tool parameters carry a description.
- **Tool Safety**: 100/100
  - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.
  - All 2 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation.
  - An AI judge read all 27 captured unit(s) of tool text and found none that tries to manipulate the model reading it.
- **Capabilities**: 60/100
  - Spec-recency check failed: implements MCP spec 2025-06-18; the latest is 2026-07-28.

**Unverified: 1 category.** A category scored 0 because we could not verify it: a data source with nothing on this package, evidence we could not reach, or a check we could not run. We only credit what we can confirm.

## Install

### How do I install the KubeDrive MCP server?

KubeDrive ships as an MCPB bundle, a single file you download and open in an MCP host that supports MCPB bundles, such as Claude Desktop. The host reads the launch command from the bundle's own manifest, so there is no command to copy. Before opening it, check the file's SHA-256 matches the one the MCP registry declares for it, 0d96927e9d9821a86e143e1c0b2d283d903705ec08d1c16ebe0161f0cc2fb0c5; the commands to do that are on this page. If it does not match, do not open it.

- Download bundle: `https://github.com/aleksskoz/kubedrive/releases/download/v0.6.6/kubedrive-0.6.6.mcpb`
- SHA-256: `0d96927e9d9821a86e143e1c0b2d283d903705ec08d1c16ebe0161f0cc2fb0c5`

### Verify on macOS / Linux

```bash
echo "0d96927e9d9821a86e143e1c0b2d283d903705ec08d1c16ebe0161f0cc2fb0c5  kubedrive-0.6.6.mcpb" | shasum -a 256 -c
```

### Verify on Windows (PowerShell)

```powershell
(Get-FileHash -Algorithm SHA256 .\kubedrive-0.6.6.mcpb).Hash -eq "0d96927e9d9821a86e143e1c0b2d283d903705ec08d1c16ebe0161f0cc2fb0c5"
```

## Changelog

Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims.

### 2026-11-04 (score 37, +1)

No change was recorded against any check on this day. Stability & Change Management went from 13 to 17. That category is still filling its 30-day observation window: 4 days of observed history at the previous scan, 5 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-11-02 (score 36, +1)

No change was recorded against any check on this day. Stability & Change Management went from 7 to 10. That category is still filling its 30-day observation window: 2 days of observed history at the previous scan, 3 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-10-31 (score 35, +1)

- [security regression] Stability: 0.00 → unverified

### 2026-10-30 (score 34)

First indexed and scored.

## MCP tools (26)

### `list_clusters` (~33 tokens)

List clusters

Kubernetes clusters (kubeconfig contexts) the user shared with agents in KubeDrive, with API server and default namespace.

### `list_resource_types` (~59 tokens)

List resource types

Resource types the cluster serves: type id, kind, whether namespaced. Use the ids with the other tools.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `filter` (string): substring of id or kind

### `list_resources` (~137 tokens)

List resources

Objects of a type, as kubectl get shows them (server-side columns), one per line.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `field_selector` (string): e.g. status.phase=Running
- `label_selector` (string): e.g. app=web,tier!=db
- `limit` (integer): default 500
- `namespace` (string): namespace; empty for all namespaces or cluster-scoped types
- `type` (string, required): resource type id: "pods", "deployments.apps", "ingresses.networking.k8s.io" (list_resource_types)

### `get_resource` (~111 tokens)

Get resource

One object as YAML, without managedFields. Secret values are removed; credential-looking fields, env values and strings are masked (best effort).

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `name` (string, required): object name
- `namespace` (string): namespace; empty for all namespaces or cluster-scoped types
- `type` (string, required): resource type id: "pods", "deployments.apps", "ingresses.networking.k8s.io" (list_resource_types)

### `get_events` (~103 tokens)

Get events

Events of one object (type and name), or of a namespace (or all namespaces), newest first.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `limit` (integer): default 200
- `name` (string): object name
- `namespace` (string): namespace; empty for all namespaces or cluster-scoped types
- `type` (string): with name: events of that object
- `warnings_only` (boolean)

### `get_logs` (~109 tokens)

Get pod logs

The last lines of a pod's container log, with timestamps. Secret-looking values are redacted.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `container` (string): needed when the pod has several
- `contains` (string): only lines containing this text
- `namespace` (string, required)
- `pod` (string, required)
- `previous` (boolean): the log of the run before the last restart
- `tail_lines` (integer): default 200

### `cluster_problems` (~59 tokens)

Cluster problems

What needs attention, as KubeDrive's cluster overview shows it: failing pods and workloads, nodes not ready, failed Helm releases, and the last hour's warning events.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters

### `top` (~76 tokens)

Resource usage

CPU (millicores) and memory of pods or nodes from metrics-server, busiest first.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `kind` (string, required)
- `limit` (integer): default 50
- `namespace` (string): namespace; empty for all namespaces or cluster-scoped types

### `check_access` (~126 tokens)

Check access

Whether the user may do a verb on a type (kubectl auth can-i), by SelfSubjectAccessReview.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `name` (string): object name
- `namespace` (string): namespace; empty for all namespaces or cluster-scoped types
- `type` (string, required): resource type id: "pods", "deployments.apps", "ingresses.networking.k8s.io" (list_resource_types)
- `verb` (string, required): get, list, watch, create, update, patch, delete…

### `diagnose` (~175 tokens)

Diagnose object

Why an object is unhealthy, by KubeDrive's rules (no AI): each finding with severity, the cause, the evidence (cluster messages, the crashed container's last log lines) and a suggested fix. Types: pods, deployments.apps, statefulsets.apps, daemonsets.apps, replicasets.apps, jobs.batch, services, ingresses.networking.k8s.io, persistentvolumeclaims, nodes. Start here before reading logs and events by hand.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `name` (string, required): object name
- `namespace` (string): namespace; empty for all namespaces or cluster-scoped types
- `type` (string, required): resource type id: "pods", "deployments.apps", "ingresses.networking.k8s.io" (list_resource_types)

### `broken_links` (~120 tokens)

Broken links

Links between objects that lead nowhere, as KubeDrive's map shows them: Ingress rules to a missing Service or port, Services that select no pods, pods that mount or read a ConfigMap, Secret or PVC that does not exist, autoscalers whose target is gone, disruption budgets that match no pods. Optional references are not judged, nor kinds the user may not list.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `namespace` (string): namespace; empty for all namespaces (slower on big clusters)

### `upgrade_check` (~118 tokens)

Upgrade check

What breaks when the cluster moves to a newer Kubernetes minor version (KubeDrive Pro): Helm releases and kubectl-applied manifests that name removed API versions, clients still calling them, kinds removed with no successor, and nodes whose kubelet is too old. Each finding says what to write instead. Read only.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `target` (integer): the minor version of 1.x to check, e.g. 33 for 1.33; default the next one

### `timeline` (~228 tokens)

Timeline

What changed before it broke: one feed, newest first, of the cluster's events, rollouts (new ReplicaSet revisions with the image that changed), container restarts with exit codes, changes made through KubeDrive with who made them, Helm revisions and firing alerts. With type and name: that object and what it is joined to; without: the whole cluster. Start here for "what happened" and "why did it break". Windows beyond 1h need KubeDrive Pro.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `linked` (boolean): include the objects it is joined to (pods, Services, Ingresses, ConfigMaps, autoscaler); default true
- `name` (string): object name
- `namespace` (string): namespace; empty for all namespaces or cluster-scoped types
- `type` (string): resource type id: "pods", "deployments.apps", "ingresses.networking.k8s.io" (list_resource_types)
- `window` (string): how far back; default 1h

### `rightsizing` (~133 tokens)

Rightsizing

Compares what the cluster's containers request with what they used, from Prometheus (KubeDrive Pro): workloads at risk (killed for memory, throttled by a CPU limit, no requests, using more than they request) and over-requested ones, each container with current and suggested requests and limits (CPU from the 95th percentile, memory from the highest use, plus the user's headroom). Read only; set resources with the edit tools.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `window` (string): how far back to look; default 7d

### `apply` (~170 tokens)

Apply a manifest

Changes an object to match the manifest (kubectl apply on one object): a server-side dry run first, then the user sees the diff in KubeDrive and allows or declines. The manifest is the whole object as get_resource returns it, edited; fields left out are removed.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `manifest` (string, required): the object's full YAML
- `name` (string, required): object name
- `namespace` (string): namespace; empty for all namespaces or cluster-scoped types
- `reason` (string): why, in one line; the user sees it with the change
- `type` (string, required): resource type id: "pods", "deployments.apps", "ingresses.networking.k8s.io" (list_resource_types)

### `scale` (~132 tokens)

Scale a workload

Sets the replicas of a Deployment, StatefulSet or ReplicaSet, after the user allows it in KubeDrive.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `name` (string, required): object name
- `namespace` (string, required): namespace; empty for all namespaces or cluster-scoped types
- `reason` (string): why, in one line; the user sees it with the change
- `replicas` (integer, required)
- `type` (string, required): resource type id: "pods", "deployments.apps", "ingresses.networking.k8s.io" (list_resource_types)

### `restart` (~128 tokens)

Restart a workload

Rolls out new pods for a Deployment, StatefulSet or DaemonSet (kubectl rollout restart), after the user allows it.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `name` (string, required): object name
- `namespace` (string, required): namespace; empty for all namespaces or cluster-scoped types
- `reason` (string): why, in one line; the user sees it with the change
- `type` (string, required): resource type id: "pods", "deployments.apps", "ingresses.networking.k8s.io" (list_resource_types)

### `rollback` (~151 tokens)

Roll back a workload

Returns a Deployment, StatefulSet or DaemonSet to an earlier revision (kubectl rollout undo); without a revision, the previous one. The user allows it in KubeDrive.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `name` (string, required): object name
- `namespace` (string, required): namespace; empty for all namespaces or cluster-scoped types
- `reason` (string): why, in one line; the user sees it with the change
- `revision` (integer): omit for the previous revision
- `type` (string, required): resource type id: "pods", "deployments.apps", "ingresses.networking.k8s.io" (list_resource_types)

### `delete` (~147 tokens)

Delete an object

Deletes one object after the user allows it in KubeDrive; namespaces and cluster-scoped objects always need the name typed by the user.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `force` (boolean): no grace period
- `name` (string, required): object name
- `namespace` (string): namespace; empty for all namespaces or cluster-scoped types
- `propagation` (string)
- `reason` (string): why, in one line; the user sees it with the change
- `type` (string, required): resource type id: "pods", "deployments.apps", "ingresses.networking.k8s.io" (list_resource_types)

### `cordon` (~64 tokens)

Cordon a node

Marks a node unschedulable, after the user allows it.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `node` (string, required)
- `reason` (string): why, in one line; the user sees it with the change

### `uncordon` (~65 tokens)

Uncordon a node

Marks a node schedulable again, after the user allows it.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `node` (string, required)
- `reason` (string): why, in one line; the user sees it with the change

### `drain` (~125 tokens)

Drain a node

Cordons a node and evicts its pods (kubectl drain; DaemonSet pods stay), after the user allows it. Pods with emptyDir volumes or without a controller are skipped unless the options say otherwise.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `delete_emptydir_data` (boolean)
- `force` (boolean): also evict pods no controller will recreate
- `node` (string, required)
- `reason` (string): why, in one line; the user sees it with the change
- `timeout_seconds` (integer)

### `trigger_cronjob` (~94 tokens)

Run a CronJob now

Creates a Job from a CronJob right away, after the user allows it; returns the Job's name.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `name` (string, required): object name
- `namespace` (string, required): namespace; empty for all namespaces or cluster-scoped types
- `reason` (string): why, in one line; the user sees it with the change

### `port_forward` (~180 tokens)

Open a port forward

Forwards a local port on this computer to a pod, service or workload, after the user allows it; returns the local port and the forward's id (port_forward_stop closes it). The forward stays open until stopped or KubeDrive quits.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `local_port` (integer): omit for a free one
- `name` (string, required): object name
- `namespace` (string, required): namespace; empty for all namespaces or cluster-scoped types
- `port` (integer, required): the port in the pod or service
- `reason` (string): why, in one line; the user sees it with the change
- `type` (string, required): resource type id: "pods", "deployments.apps", "ingresses.networking.k8s.io" (list_resource_types)

### `port_forward_stop` (~33 tokens)

Close a port forward

Closes a port forward opened by port_forward, by its id. No confirmation.

Input parameters:

- `id` (string, required)

### `exec` (~150 tokens)

Run a command in a container

Runs one command (an argument list, no shell) in a container without a terminal, after the user allows it: stdout, stderr and the exit code come back, each output cut at 256 KB. Up to 60 seconds.

Input parameters:

- `cluster` (string, required): kubeconfig context name, from list_clusters
- `command` (array, required): the program and its arguments, e.g. ["nginx","-T"]
- `container` (string): needed when the pod has several
- `namespace` (string, required): namespace; empty for all namespaces or cluster-scoped types
- `pod` (string, required)
- `reason` (string): why, in one line; the user sees it with the change

## Diagnostics

Captured diagnostic sections: Provenance. The full working is on the page: https://verifymcp.io/servers/aleksskoz-kubedrive/kubedrive#diagnostics

## Score history

- 2026-11-05: 37
- 2026-11-04: 37
- 2026-11-03: 36
- 2026-11-02: 36
- 2026-11-01: 35
- 2026-10-31: 35
- 2026-10-30: 34

## Common questions

### What is the KubeDrive MCP server?

KubeDrive is an MCP server listed in the public MCP registry as io.github.aleksskoz/kubedrive. Your Kubernetes clusters for AI agents through the KubeDrive desktop app, without a kubeconfig. This page covers its MCPB bundle (https://github.com/aleksskoz/kubedrive/releases/download/v0.6.6/kubedrive-0.6.6.mcpb).

### Is the KubeDrive MCP server safe to use?

KubeDrive scores 37 out of 100 on VerifyMCP. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

### What tools does the KubeDrive MCP server expose?

KubeDrive exposes 26 tools: list_clusters, list_resource_types, list_resources, get_resource, get_events, and 21 more. Their descriptions and schemas cost roughly 3,026 tokens of context every time the server is loaded.

## Links

- Repository: https://github.com/aleksskoz/kubedrive
- Website: https://kubedrive.com/docs/ai-agents
- Changelog RSS feed: https://verifymcp.io/servers/aleksskoz-kubedrive/kubedrive.xml
- Changelog JSON feed: https://verifymcp.io/servers/aleksskoz-kubedrive/kubedrive.json
- HTML version of this page: https://verifymcp.io/servers/aleksskoz-kubedrive/kubedrive