> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ankra.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Introduction
> Ankra is one Kubernetes platform with four first-class surfaces: a web UI with Stack Builder and AI Assistant, a scriptable CLI (ankra chat, ankra cluster apply), a remote MCP server with 169 tools for Claude/Cursor/agents, and CI/CD integrations (GitHub Actions, GitLab, pipeline agents, Terraform). One tool registry powers them all.
**Are you an AI agent or automating Ankra?** Don't drive the web UI with a
browser. Connect over the [MCP server](/platform/mcp-server) (recommended) or
the [CLI](/integrations/ankra-cli) - both give you the same tools the product's
own AI uses, with scoped read/write access and a full audit trail. Your CI jobs
can call the CLI directly too - see [Pipeline Agents](/guides/pipeline-agents).
Setup is in [Connect an AI agent](#connect-an-ai-agent) below.
## What is Ankra?
Ankra is one platform to build, deploy, and debug Kubernetes - operated from whichever surface fits the job: the **web UI** for humans, the **CLI** for terminals and scripts, an **AI agent over MCP** for Claude, Cursor, and other MCP clients, or your **CI/CD pipelines**.
Import any cluster - EKS, GKE, AKS, on-prem, or local - and manage it from one place. Or provision managed clusters on Hetzner, OVHCloud, or UpCloud directly from the platform.
All four surfaces share one tool registry: the same 169 tools power the in-product AI Assistant, the MCP server, chat over Slack and Microsoft Teams, and SCM comments - so what you can ask Ankra to do is the same everywhere.
***
## Four Ways to Operate Ankra
The visual Stack Builder, the `⌘+J` AI Assistant with Ask, Agentic, and Plan modes, and a full Kubernetes resource browser. Built for humans.
`ankra chat "..." --mode ask|agent` for one-shot AI turns, `ankra cluster apply -f` for declarative cluster operations. Built for terminals, scripts, and CI.
A remote MCP server exposing 169 tools with `mcp:read`/`mcp:write` scopes and OAuth. Built for Claude, Cursor, and any other MCP client.
GitHub Actions and GitLab image-bump pipelines, Pipeline Agents (plain CI jobs that call the CLI), and a Terraform provider. Built for your delivery workflows.
***
## Connect an AI agent
Ankra is built to be operated by agents, not just people. **No browser needed:
don't scrape the platform or automate the web UI** - connect over MCP or the CLI
instead.
**Fastest path for Claude Code** - register the MCP server once, then authenticate
with `/mcp`:
```bash theme={null}
claude mcp add --transport http ankra https://platform.ankra.app/api/v1/mcp
```
For Claude desktop/web add Ankra as a custom connector; for Cursor drop the
endpoint into your MCP config. Full setup for every client, token scopes, and the
OAuth flow are in the [MCP Server guide](/platform/mcp-server).
Then teach your editor's agent Ankra's conventions - `ankra skills install`
installs Ankra's curated Agent Skills into Cursor (default) or Claude Code
(`--editor claude-code`).
Every tool the MCP server exposes, the scope it needs, and whether it targets
a specific cluster.
Teach Ankra's AI your conventions with organisation-scoped skills that
override the platform defaults.
Plain CI jobs that call `ankra chat` and `ankra cluster apply` - PR review,
deploy watching, health checks. No framework to run.
Ankra's autonomous infrastructure agent, operated from Slack, Discord, or
Telegram chat.
Start read-only (`mcp:read`) to explore, then grant `mcp:write` when the agent
should apply changes. Every resolved tool call - including refused ones - is
written to the organisation audit log.
***
## Build Your Stack in the UI
Stop copy-pasting YAML between clusters. The Stack Builder gives you a visual canvas to compose Kubernetes environments that are reusable, versioned, and deployed consistently.
* **Visual drag-and-drop** - add Helm charts and manifests, define dependencies, configure values
* **AI-assisted composition** - describe what you need ("monitoring stack with Prometheus, Grafana, and alerting") and the AI recommends components, configurations, and deployment order
* **Variables** - parameterise manifests with organisation, cluster, and stack-level variables using `${{ ankra.variable_name }}` syntax
* **Clone across clusters** - replicate a stack from dev to staging to production in seconds
* **SOPS encryption** - encrypt secrets in your GitOps repository with built-in SOPS support
***
## Automate Through GitOps and CI/CD
Ankra's GitOps engine is designed to slot into your existing workflows, not replace them. Connect a GitHub repository, and your cluster configuration becomes code - versioned, reviewable, and automatically applied.
### Works with existing pipelines
Your CI builds and pushes container images. A commit to the GitOps repo updates the image tag. Ankra detects the change and deploys. No new tooling required.
```mermaid theme={null}
flowchart LR
A[Code Push] --> B[CI builds image]
B --> C[Update GitOps repo]
C --> D[Ankra syncs cluster]
D --> E[Deployed]
```
### Modular repository structure
Split your configuration into files and folders with `include` paths. Different teams own different files. No merge conflicts, no giant config files.
```yaml theme={null}
apiVersion: v1
kind: ImportCluster
metadata:
name: my-cluster
spec:
git_repository:
provider: github
credential_name: my-credential
branch: main
repository: my-org/my-repo
stacks:
- name: platform-stack
manifests:
- include: manifests/
addons:
- include: addons/
```
### Full audit trail
Every sync is tracked - commit SHA, trigger source, status, and timing. View sync history, monitor progress, and debug failures from the platform or via the API.
### Plug in your pipelines
The canonical build → push → image-bump → sync pipeline for GitHub repositories.
The same pipeline for GitLab CI, using the GitLab Container Registry and CI variables.
Turn plain CI jobs into infrastructure agents with `ankra chat` and `ankra cluster apply --wait`.
Repo in, generated Dockerfile, Helm chart, and CI/CD out - shipped as a stack.
Prefer infrastructure-as-code? The [Terraform provider](/integrations/terraform) manages clusters, stacks, manifests, and add-ons declaratively.
***
## Manage and Debug with AI
Ankra's AI isn't a generic chatbot bolted onto a dashboard. It has full context - your pod logs, Kubernetes manifests, stack deployments, resource states, events, and their relationships. This makes it exceptionally good at incident triangulation.
### AI Assistant
Press `⌘+J` from anywhere. Choose **Ask** (read-only), **Agentic** (confirm each change), or **Plan** (approve once, then execute). The AI is page-aware - open a crashing pod and it already sees the logs, events, and manifest. Ask it anything:
* *"Why is this pod failing?"* - correlates logs, events, and resource state
* *"Create a monitoring stack with Prometheus and Grafana"* - designs and deploys a complete stack
* *"What changed in the last hour that could cause this?"* - cross-references deployment history with symptoms
### Proactive AI Insights
AI Insights scan your clusters on a schedule and surface issues before they become incidents:
* Root cause analysis with remediation commands
* Severity tracking and trend analytics (MTTR, category breakdowns)
* Adaptive scanning - critical issues scanned every 60s, healthy clusters every 10 minutes
### AI Incidents
When alerts trigger, the AI automatically collects pod status, events, logs, and node state - then delivers a structured analysis with root cause, affected resources, and recommended actions.
***
## Everything Else
Browse pods, deployments, services, secrets, RBAC, storage, CRDs, and Helm releases. Tail logs. Edit configs. No kubectl required.
Configure alerts with AI-powered incident analysis. Send notifications to Slack, webhooks, or any external system.
Provision K3s clusters on Hetzner, OVHCloud, or UpCloud with automated networking, bastion access, and NAT gateways.
Estimated spend per cluster and across your fleet, broken down by compute, storage, and namespace.
Capture a working stack as a reusable, versioned, parameterised template and roll it out anywhere.
Give teammates scoped, SSO-backed `kubectl` access through Ankra's proxy - even to private clusters.
A world map of every cluster with health and cost rollups across your whole organisation.
Full REST API for programmatic access to clusters, stacks, insights, and operations.
***
## Quick Start
Create an account at [platform.ankra.app](https://platform.ankra.app)
Import an existing cluster by installing the Ankra agent, or provision a new one on Hetzner, OVHCloud, or UpCloud.
```bash theme={null}
helm upgrade --install ankra-agent oci://ghcr.io/ankraio/ankra-agent/ankra-agent \
--namespace ankra \
--create-namespace \
--set config.token="YOUR_UNIQUE_TOKEN"
```
Open the Stack Builder or press `⌘+J` and tell the AI what you need:
*"Set up a monitoring stack with Prometheus, Grafana, and Loki"*
Link a GitHub repository in cluster settings. Your stacks are now version-controlled and automatically synced.
[Full Getting Started Guide →](/get-started/quickstart)
***
## Next Steps
Step-by-step guide from first cluster to full GitOps pipeline.
Turn plain CI jobs into infrastructure agents with the Ankra CLI.
Get help and share feedback.