Exposes the Ollama SDK as MCP tools so MCP clients can use local LLM models.
Setup
Choose your client, copy the config, then fill in any required environment variables.
claude mcp add ollama -- npx -y ollama-mcp~/.codex/config.toml
[mcp_servers.ollama]
command = "npx"
args = ["-y", "ollama-mcp"]
# codex mcp add ollama -- npx -y ollama-mcp~/.cursor/mcp.json
{
"mcpServers": {
"ollama": {
"command": "npx",
"args": [
"-y",
"ollama-mcp"
]
}
}
}.vscode/mcp.json
{
"servers": {
"ollama": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"ollama-mcp"
]
}
}
}~/.gemini/settings.json
{
"mcpServers": {
"ollama": {
"command": "npx",
"args": [
"-y",
"ollama-mcp"
]
}
}
}claude_desktop_config.json
{
"mcpServers": {
"ollama": {
"command": "npx",
"args": [
"-y",
"ollama-mcp"
]
}
}
}.trae/mcp.json
{
"mcpServers": {
"ollama": {
"command": "npx",
"args": [
"-y",
"ollama-mcp"
]
}
}
}