OKE MCP Server
MCP server for Oracle Container Engine for Kubernetes (OKE) that enables inspection, querying, and troubleshooting of OKE clusters through safe, composable tools.
README
OKE MCP Server
Model Context Protocol (MCP) server for Oracle Container Engine for Kubernetes (OKE). It lets MCP‑aware chat clients (e.g. Claude Desktop, VS Code Agent, custom CLI) inspect, query and troubleshoot your OKE clusters through a small set of safe, composable tools.

✨ Highlights
- Lean, LLM‑friendly APIs – small, consistent payloads (
{items,next,error}/{item,error}) with optionalverboseandhints. - OCI auth that “just works” – supports security token (local dev) and API keys.
- No venv required – run with uvx:
uvx oke-mcp-server --transport stdio. - Token rotation – refresh without restart using the
auth_refreshtool. - Production‑ready ergonomics – clear errors, pagination, predictable shapes.
Requirements
- Python 3.10+ (3.11+ recommended)
- uv (recommended): https://github.com/astral-sh/uv
- OCI credentials configured locally (see below)
Optional:
- MCP host/client (Claude Desktop, VS Code Agent Mode, etc.)
Install & Run (recommended)
Run the published package with uvx:
uvx oke-mcp-server --transport stdio
Or run a specific version:
uvx --from oke-mcp-server==0.2.* oke-mcp-server --transport stdio
The server speaks MCP over stdio. Most MCP hosts handle initialization automatically. For raw testing you can still send JSON‑RPC (see “Quick JSON‑RPC test”).
Configure Authentication
Option A — Security Token (best for local dev)
- Sign in via the OCI Console/CLI to obtain a security token.
- In your
~/.oci/configprofile (e.g.DEFAULT) include:[DEFAULT] tenancy=ocid1.tenancy.oc1..aaaa... region=eu-frankfurt-1 user=ocid1.user.oc1..aaaa... # usually present; not used by STS signer key_file=/path/to/your/api_key.pem # keep if you also use API key flows fingerprint=XX:XX:... # same as above security_token_file=/path/to/token # REQUIRED for STS - Export (or set in your MCP host env):
export OCI_CLI_AUTH=security_token - (When the token expires) call the MCP tool:
{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"auth_refresh","arguments":{}}}
Option B — API Key
Use your standard ~/.oci/config profile with user, key_file, fingerprint, tenancy, region. Do not set OCI_CLI_AUTH=security_token.
The server also honors
OKE_COMPARTMENT_IDandOKE_CLUSTER_IDenvironment variables as defaults.
Using with an MCP Host
Claude Desktop (example)
Settings → MCP Servers → Add:
{
"name": "oke",
"command": "uvx",
"args": ["oke-mcp-server", "--transport", "stdio"],
"env": {
"OCI_CLI_AUTH": "security_token",
"OKE_COMPARTMENT_ID": "ocid1.compartment.oc1..aaaa...",
"OKE_CLUSTER_ID": "ocid1.cluster.oc1.eu-frankfurt-1.aaaa..."
}
}
That’s it—Claude will list the tools and can call them during chat.
Quick JSON‑RPC test (manual)
Start the server:
uvx oke-mcp-server --transport stdio
Then send:
{"jsonrpc":"2.0","id":0,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"manual","version":"0.0.0"}}}
{"jsonrpc":"2.0","method":"notifications/initialized","params":{}}
{"jsonrpc":"2.0","id":1,"method":"tools/list"}
{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":"meta_health","arguments":{}}}
Tools (stable set)
All list tools return:
{ "items": [...], "next": "<token|null>", "error": null, "meta": { ... } }
Single‑item tools return:
{ "item": { ... }, "error": null, "meta": { ... } }
Common inputs:
limit(default 20),continue_token(pagination)verbose: bool(include extra details)hints: bool(include lightweight graph hints where applicable)auth: "security_token" | null(override; otherwise server uses env/defaults)
Meta / Config
- meta_health →
{server, version, defaults, effective} - meta_env → redacted env snapshot for diagnostics
- auth_refresh → re‑loads auth (use after rotating security token)
- config_get_effective_defaults / config_set_defaults → manage fallback OCIDs
OKE / Kubernetes
- k8s_list — list Pods, Services, Namespaces, Nodes, Deployments, ReplicaSets, Endpoints, EndpointSlices, Ingress, Gateway, HTTPRoute, PVC, PV, StorageClass
- k8s_get — get a single resource by kind/namespace/name
- oke_get_pod_logs — stream recent logs from a container (supports
tail_lines,since_seconds,previous,timestamps) - oke_list_clusters / oke_get_cluster — cluster discovery and details (OCI)
For public logs on OKE, ensure worker nodes allow the API->kubelet path: TCP/10250 from the control‑plane CIDR/NSG. Timeouts when calling
read_namespaced_pod_logtypically mean this network path is blocked.
Troubleshooting
| Symptom | Likely Cause | Fix |
|---|---|---|
{'user':'missing'} from OCI SDK |
No valid signer or profile | Set OCI_CLI_AUTH=security_token or ensure ~/.oci/config has user/key_file/fingerprint |
| TLS bundle not found | Wrong Python cert path | Ensure certifi is installed in the environment running the server |
Logs 500 / i/o timeout to :10250 |
Control‑plane → node kubelet blocked | Open TCP/10250 from API endpoint CIDR in Security List / NSG |
Tool says cluster_id required |
No defaults present | Set OKE_CLUSTER_ID env or call config_set_defaults |
Project Structure
oke_mcp_server/
__init__.py
main.py
auth.py
config_store.py
tools/
k8s.py
oke_cluster.py
pyproject.toml
Makefile
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.