Right now the MCP server can read schema information with getSchema, but can't modify schemas or solrconfig.xml. This means users have to manually edit XML files to set up things like vector search fields or codec factories, then come back to use the MCP tools. It breaks the workflow of managing collections entirely through AI assistants.

It would be helpful to add a few MCP tools: addFieldType to create field types (like DenseVectorField with all its parameters), addField to add fields using existing field types, and addCodecFactory to add codec factories to solrconfig.xml. This would let users set up collections end-to-end via prompts instead of switching between the MCP server and manual config editing.

Implementation would use SolrJ's existing SchemaRequest.AddFieldType, SchemaRequest.AddField, and ConfigRequest APIs, following the same pattern as the current getSchema tool.