docs: Added pointer to mcp docs (#201)

Author: droot

README.md

@@ -181,6 +181,10 @@ kubectl-ai --llm-provider=openai --model=gpt-4.1 --quiet "scale the nginx deploy
 
 The `kubectl-ai` will process your query, execute the appropriate kubectl commands, and provide you with the results and explanations.
 
+## MCP server
+
+You can also use `kubectl-ai` as a MCP server that exposes `kubectl` as one of the tools to interact with locally configured k8s environment. See [mcp docs](./docs/mcp.md) for more details.
+
 ## k8s-bench
 
 kubectl-ai project includes [k8s-bench](./k8s-bench/README.md) - a benchmark to evaluate performance of different LLM models on kubernetes related tasks. Here is a summary from our last run:

docs/mcp.md

@@ -1,6 +1,6 @@
 # `kubectl-ai` as a MCP Server
 
-`kubectl-ai` can ack as a [Model Context Protocol (MCP)](https://github.com/modelcontextprotocol) server. This allows AI agents and IDEs to act as MCP clients and connect to the `kubectl-ai` MCP server, effectively interacting with your local Kubernetes environment via `kubectl`.
+`kubectl-ai` can act as a [Model Context Protocol (MCP)](https://github.com/modelcontextprotocol) server. This allows MCP clients such as AI agents and IDEs (Claude, Cursor) to connect to the `kubectl-ai` MCP server, effectively interacting with your local Kubernetes environment via `kubectl`.
 
 ## Overview
 
@@ -70,4 +70,4 @@ Cursor also supports MCP servers. You can [configure `kubectl-ai` MCP Server](ht
     *   Refer to the documentation of your specific MCP client (Claude, Cursor, etc.) for troubleshooting steps related to their MCP implementation.
     *   Check the client's logs for any error messages related to MCP communication.
 *   Error Logs:
-    *   `kubectl-ai` logs errors. If it's launched via an MCP client, refer to the documentation of your specific MCP client to see how to access the logs directly. You can also run `kubectl-ai --mcp-server -v=X` manually (where X is a log level) to see if it reports any issues when a client tries to connect or send a command.
+    *   `kubectl-ai` logs errors. If it's launched via an MCP client, refer to the documentation of your specific MCP client to see how to access the logs directly. You can also run `kubectl-ai --mcp-server -v=X` manually (where X is a log level) to see if it reports any issues when a client tries to connect or send a command.