Tekton Toolset

The tekton toolset adds Tekton-specific helpers on top of the generic Kubernetes resource tools.

Enable it with:

kubernetes-mcp-server --toolsets core,config,tekton

PipelineRun operations

• tekton_pipeline_start starts a Pipeline by creating a PipelineRun.
• tekton_pipelinerun_lifecycle restarts a PipelineRun from its existing spec or cancels it by setting spec.status to Cancelled.
• tekton_pipelinerun_logs collects logs from TaskRuns owned by a PipelineRun. Use the optional task and step parameters to narrow the output.

TaskRun operations

• tekton_task_start starts a Task by creating a TaskRun.
• tekton_taskrun_restart creates a new TaskRun from an existing TaskRun spec.
• tekton_taskrun_logs resolves a TaskRun pod and returns step/sidecar logs. Use the optional step parameter to return one step only.

Pipeline-as-Code Repository and operator TektonConfig resources are ordinary Kubernetes resources; use resources_list and resources_get for those.

List Pipeline-as-Code repositories in a namespace with resources_list:

{"apiVersion":"pipelinesascode.tekton.dev/v1alpha1","kind":"Repository","namespace":"my-namespace"}

Get the usual cluster TektonConfig with resources_get:

{"apiVersion":"operator.tekton.dev/v1alpha1","kind":"TektonConfig","name":"config"}

These tools are read-only except the start and lifecycle operations.

Troubleshooting prompt

Use the pipeline-troubleshoot prompt with namespace and name to gather PipelineRun status, the exact resolved or embedded PipelineSpec when available, the referenced Pipeline otherwise, related TaskRuns, failed or errored step logs, warning events, Pipeline-as-Code repositories, and TektonConfig context into one diagnostic prompt. Missing definitions are reported without preventing the remaining data from being collected. The guide structures the response as PipelineRun Status, TaskRun Status, Failed Step Logs, Events, Troubleshooting Analysis, and Fix Suggestions.