‹ Back to apps
Ankra
Build, ship and operate Kubernetes with Ankra: the hosted Ankra MCP server, 25 agent skills for clusters, stacks, GitOps, CI/CD, secrets, observability, security and troubleshooting, and slash commands to stand up a cluster, ship a service and triage a failure.
What’s inside
1 MCP server · 25 skills- MCP serverankraMCP server: ankra
- Skillankra-ai-agentsRun and govern Ankra's AI agents - choosing the model provider and catalogue (Ankra, Anthropic, OpenRouter, or an OpenAI-compatible endpoint such as LiteLLM or vLLM), registering MCP tool servers and granting individual tools to organisation roles, dispatching and supervising agent runs and their transcripts, working the AI board of tickets, and confirming or rejecting the writes an agent proposes. Use when the user mentions Ankra AI, AI agents, sub-agents, agent runs, MCP servers or tools, the AI board or tickets, the model catalogue, or configuring which model Ankra uses.
- Skillankra-ai-gatewayConfigure Ankra's AI Gateway across Slack, Microsoft Teams, GitHub, GitLab, and Bitbucket, set per-binding Ask/Agent safety modes, pick the org staging cluster for ephemeral AI workspace pods and PR demos, and run the pipeline-failure investigation and auto-fix-PR flow. Use when the user connects an AI integration, mentions Ask vs Agent mode, AI workspaces, PR demos, a staging cluster for AI, or wants the AI to investigate a failing pipeline and open a fix PR.
- Skillankra-alerts-webhooksRoute Ankra's platform notifications to Slack, Microsoft Teams, Discord, PagerDuty or a custom webhook - creating destinations (webhook URLs or bot channels), routing rules that filter by kind, severity and cluster with include/exclude modes and priorities, previewing which destinations a notification would reach, and testing delivery before relying on it. Use when the user wants alerting, notifications, incident routing, on-call integration, webhook delivery from Ankra, or asks why a notification did or did not reach a channel.
- Skillankra-app-integrationsConnect an application to services and secrets that already exist in an Ankra organisation - an LLM gateway such as LiteLLM or vLLM, a container registry such as Harbor, a database, an object store, or an internal API - by reusing stored credentials rather than minting new ones, and choosing correctly between application env-secrets, SOPS-encrypted manifests, and cluster/organisation variables. Use when the user wants to wire an app to LiteLLM, Harbor, Postgres, Redis, S3 or an internal service, mentions connecting to an existing secret or app, or asks where a configuration value should live.
- Skillankra-applicationsTake source code from a Git repository to a running deployment on one or many Ankra clusters - registering an application, the generated Dockerfile, chart and `.ankra/pipeline.yaml` (Ankra Pipelines, never a GitHub Actions workflow), the image registry it publishes to, environment secrets, deployments, auto-deploy, PR demos, and publishing the result as a catalogue add-on. Use when the user wants to deploy their own code with Ankra, mentions `ankra application`, connects a repository, asks how to get a service live, or wants the same service on several clusters.
- Skillankra-backupsManage the organisation's backup vaults with `ankra backup vaults` - the S3-compatible buckets cluster backups and migration data move through - and, where the closed-beta `backups` feature is on, the restore points over them: protect a stack on a schedule, take a restore point now, inspect what one carries and what it does not, restore a stack in place, clone a stack onto another cluster with its data, and follow the runs that move the data. Use when the user mentions backups, backup vaults, restore points, protecting or restoring a stack, cloning a stack with its data, object storage for backups, an S3 or MinIO bucket for the platform, or asks where migration dumps are kept.
- Skillankra-cicdPut any repository's CI/CD on Ankra Pipelines - a committed `.ankra/pipeline.yaml` that Ankra runs in-cluster on the organisation's CI cluster, for application repositories (`ankra application add` lands on it) AND for bare repositories with no application (platform code, libraries, docs, values repos) via `ankra pipeline repositories connect` - with the authoring primer, the `Ankra pipeline` check run, the human-only authority approval gate, and the GitOps-bump deploy pattern. Use whenever the user wants CI, a pipeline, tests on every PR or push, build-and-push on merge, or asks how to deploy on push; whenever GitHub Actions is failing, out of budget ("an Actions budget is preventing further use"), slow, or should be replaced; for a repository without an application; and whenever an Ankra pipeline produced no run, a run was skipped with a fatal violation, or a build died on a rootlesskit mount error. Never write a `.github/workflows` or `.gitlab-ci.yml` CI file for a repository Ankra can reach - that manual pattern is only for repositories on a provider Ankra cannot connect.
- Skillankra-cliDrive the Ankra CLI end to end - log in, switch organisation and cluster, apply cluster/stack YAML, manage applications and stack profiles, inspect Kubernetes resources, logs, events and metrics, triage operations, work the AI board, and script it all with `-o json` and stable exit codes. Use when the user mentions the `ankra` CLI, `ankra login`, `ankra cluster`, `ankra application`, `ankra stack-profiles`, `ankra tickets`, applying an ImportCluster, or managing an Ankra-managed cluster from the terminal.
- Skillankra-cloud-clustersProvision and operate clusters Ankra builds on cloud infrastructure - Hetzner, OVHcloud, UpCloud, DigitalOcean, Scaleway, AWS (self-managed EC2), Proxmox VE and HPE Morpheus - choosing the region and the right instance family, picking kubeadm (the default) or k3s and the etcd topology, wiring the generated stack straight into a GitOps repository, taking the ingress/DNS/TLS batteries at create time, then scaling, node groups, availability zones, upgrades and teardown. Use when the user wants Ankra to build a cluster rather than import one, asks which server type or region to pick, or mentions Hetzner, OVH, UpCloud, DigitalOcean, Scaleway, AWS EC2 (not EKS), Proxmox or Morpheus clusters.
- Skillankra-domains-dnsGive Ankra clusters public hostnames with DNS and TLS - the generated per-cluster subdomain under ankra.cc, registering your own root domain delegated to Ankra, publishing records in the organisation's zone, serving zones you hold in your own DNS account with your own credential (org-wide or per cluster), and the separate preview domain PR demos publish under. Use when the user wants a custom domain, asks why an ingress hostname does not resolve or has no certificate, mentions external-dns, DNS records, delegation, or where preview and demo URLs come from.
- Skillankra-getting-startedOnboard from an empty Ankra organisation to a running application - installing and authenticating the CLI, storing cloud/Git/registry credentials, deciding between importing a cluster and letting Ankra build one, wiring the GitOps repository, choosing the domain and DNS, laying down the base platform stack, and deploying the first app - in the order that avoids rework. Use when the user is new to Ankra, is setting up an organisation or their first cluster, asks "how do I start" or "what do I do first", or needs the end-to-end onboarding path rather than one subsystem.
- Skillankra-gitopsStructure an Ankra GitOps repository with modular include paths so cluster and stack definitions are stored in Git and synced to the cluster by the Ankra engine. Use when the user connects a Git repository to a cluster, organizes a GitOps repo layout, or wants Git to be the source of truth for their Kubernetes configuration.
- Skillankra-helm-registriesConnect HTTP and OCI Helm chart registries to Ankra and store least-privilege registry credentials (Harbor, Nexus, JFrog Artifactory, ChartMuseum, GHCR, Amazon ECR, Google Artifact Registry, Azure Container Registry, Docker Hub). Use when the user needs private charts, mentions Helm registries, OCI registries, or registry credentials in Ankra.
- Skillankra-import-clusterAuthor and apply an Ankra ImportCluster YAML manifest that connects an existing Kubernetes cluster, wires a GitOps repository, and declares stacks of manifests and Helm addons with dependency ordering. Use when the user wants to import or onboard a cluster into Ankra, write a cluster.yaml / ImportCluster file, or apply one with `ankra cluster apply`.
- Skillankra-managed-kubernetesProvision, import, and operate provider-managed Kubernetes through Ankra - DigitalOcean DOKS, UpCloud UKS, Google GKE, OVHcloud MKS, Azure AKS, Amazon EKS and Scaleway Kapsule - creating a cluster with its first node pool and GitOps wiring, discovering and adopting clusters that already exist, node pools and autoscaling, version upgrades, and deletion. Use when the user wants a managed Kubernetes cluster (the provider runs the control plane), mentions DOKS, UKS, GKE, AKS, EKS, OVH MKS or Kapsule, or wants to import a managed cluster into Ankra.
- Skillankra-migrateMove an existing deployment - docker-compose, a bare Dockerfile, or containers on a running Docker daemon - into an Ankra-managed cluster with `ankra migrate`. Plan first, convert to a stack, deploy, dump every database and restore it through a backup vault, then cut over; or run the steps separately, clean up the imports a vault keeps, and add conversion modules for other formats. Use when the user wants to migrate or move an app, VM or server deployment onto Kubernetes, mentions `ankra migrate`, docker-compose to Kubernetes, lift-and-shift, a cutover, or carrying data into a cluster.
- Skillankra-observabilityDeploy a monitoring and logging stack on an Ankra cluster (Prometheus, Grafana, Loki, Promtail) and wire external Prometheus-compatible metrics sources (Grafana Cloud, Amazon Managed Prometheus, Google Cloud Managed Prometheus, Thanos, VictoriaMetrics). Use when the user wants monitoring, metrics, dashboards, logging, or to connect a metrics source to Ankra.
- Skillankra-platform-principlesCross-cutting best practices for building and operating Kubernetes with Ankra - Git as source of truth, pinned versions, variables over hardcoding, least-privilege credentials, promotion through environments, idempotent operations, explicit confirmation for destructive actions - plus the routing map from a request to the right Ankra skill. Use whenever designing, reviewing, or changing Ankra clusters, stacks, applications, GitOps repos, or CI/CD that touches Ankra, and to decide which other ankra-* skill applies.
- Skillankra-securitySecure an Ankra organisation end to end - API tokens and their MCP scopes, organisation roles and membership, cluster access grants through the kube gateway, credential scope for Git/registry/cloud, secret handling with SOPS, application code and container scanning findings, the Security Center's fleet-wide CVE findings with CISA KEV and EPSS exploitation intelligence, and how much autonomy the AI agents and MCP tool servers are given. Use when the user asks about permissions, RBAC, who can access a cluster, token scopes, least privilege, a security review, hardening, handling secrets safely, vulnerabilities, a CVE, or what is exploited in the wild.
- Skillankra-shipShip code with Ankra - take a fresh (vibe-coded) project or any repository to a live URL through Ankra Pipelines, then run it on day 2 - the generated `.ankra/pipeline.yaml` that tests, builds rootless in your own cluster, scans with Semgrep, Checkov and Trivy, gates on findings and publishes an immutable `sha-` tag; push-to-deploy and the `Live:` URL; pull-request previews, environment secrets, rollback, promotion, alerts; and Ankra AI reviewing every pull request, watching runs and opening fix PRs. Use when the user says ship it, deploy this, get it live, go to production, wants CI/CD for a new or AI-generated project, asks how to test, build, scan, package or release code with Ankra, or asks what to do once it is live. This is the default skill for shipping; `ankra-applications` and `ankra-cicd` are its detail.
- Skillankra-sops-secretsEncrypt Kubernetes Secrets and sensitive addon values with SOPS and AGE so they can live in an Ankra GitOps repo, using `ankra cluster encrypt`/`decrypt` in cluster mode or file mode, glob key patterns that keep covering keys added later, `--all-data` for a whole Secret, and `encrypted_paths` so Ankra decrypts at deploy time. Use when the user needs to store a secret in Git, mentions SOPS, AGE, `encrypted_paths`, `ankra cluster encrypt`/`decrypt`/`sops-config`, or asks how to change a secret value without committing plaintext.
- Skillankra-stack-profilesTurn a working cluster stack into a reusable, parameterised stack profile and launch it across a fleet - builder drafts, parameters and their annotations, option sets that move several inputs together, secret parameters, publishing versions, applying to a cluster as a reviewable draft, diffing versions, sharing with other organisations, and exporting as ClusterInfrastructureAsCode YAML. Use when the user mentions stack profiles, `ankra stack-profiles`, a profile catalogue, reusing a stack across clusters, or wants one definition to serve many environments.
- Skillankra-stacks-addonsCompose Ankra Stacks from Helm addons, raw Kubernetes manifests, variables, and dependency edges that control deployment order. Use when the user is designing or editing an Ankra stack, adding a Helm addon, ordering resources with parents, or factoring configuration into variables.
- Skillankra-terraformManage Ankra clusters with the Ankra Terraform provider (ankraio/ankra 0.2.x) - provider token configuration, creating clusters on Hetzner, DigitalOcean, OVHcloud, UpCloud and Scaleway through Ankra, registering an existing cluster with its Stacks (ankra_cluster), the ankra_clusters data source, import, readiness waits, known limits, and upgrading from 0.1.x without replacing clusters. Use when the user wants to manage Ankra with Terraform, mentions the Ankra provider, `ankra_cluster` or an `ankra_*_cluster` resource, or wants infrastructure-as-code for their Ankra clusters.
- Skillankra-troubleshootingDiagnose a failing workload, deployment or Ankra operation on a managed cluster - reading logs (including the previous container of a CrashLoopBackOff), events scoped to one object, describe output, live CPU/memory from the metrics API, PromQL against the cluster's Prometheus, and the platform execution history that explains most deploy failures. Use when something is broken, a pod is crashlooping or Pending, a deploy failed, a stack will not come up, or the user asks to check logs, events, metrics or operations.