Getting Pod Logs via MCP Returns Huge Output, Lacking Filtering

[sergeykad]

Using kubectl with grep allows for filtering and reading only relevant log lines, such as errors and warnings. However, fetching logs through the MCP returns the entire log output, which can easily exceed 10,000 tokens.
We need a way to filter logs directly within the MCP.

[ardaguclu]

I think we can limit the response via https://github.com/kubernetes/kubectl/blob/6238e394935b306d29d20896dc4aaf73d2df9840/pkg/cmd/logs/logs.go#L197 or tail by default instead of retrieving full logs

[ardaguclu]

/cc @manusa