import type { Tool } from "@modelcontextprotocol/sdk/types.js";
/**
* MCP Tool definitions for Ollama
*/
export const tools: Tool[] = [
{
name: "ollama_generate",
description:
"Generate text using an Ollama model. Use this for single-turn text generation. Supports vision models (llava, llama3.2-vision, deepseek-ocr) with image input.",
inputSchema: {
type: "object",
properties: {
model: {
type: "string",
description:
"The model to use (e.g., 'llama3.2', 'codellama', 'mistral', 'llava', 'deepseek-ocr')",
},
prompt: {
type: "string",
description: "The prompt to generate from",
},
system: {
type: "string",
description: "Optional system prompt",
},
temperature: {
type: "number",
description: "Temperature for generation (0.0-2.0, default: 0.8)",
},
max_tokens: {
type: "number",
description: "Maximum tokens to generate",
},
images: {
type: "array",
items: { type: "string" },
description:
"Array of image file paths or base64-encoded images for vision models",
},
},
required: ["model", "prompt"],
},
},
{
name: "ollama_chat",
description:
"Have a multi-turn chat conversation with an Ollama model. Supports vision models (llava, llama3.2-vision, deepseek-ocr) with image input.",
inputSchema: {
type: "object",
properties: {
model: {
type: "string",
description:
"The model to use (e.g., 'llama3.2', 'codellama', 'mistral', 'llava', 'deepseek-ocr')",
},
messages: {
type: "array",
items: {
type: "object",
properties: {
role: {
type: "string",
enum: ["system", "user", "assistant"],
},
content: {
type: "string",
},
images: {
type: "array",
items: { type: "string" },
description:
"Array of image file paths or base64-encoded images",
},
},
required: ["role", "content"],
},
description: "Array of chat messages",
},
temperature: {
type: "number",
description: "Temperature for generation (0.0-2.0, default: 0.8)",
},
},
required: ["model", "messages"],
},
},
{
name: "ollama_list",
description: "List all locally available Ollama models.",
inputSchema: {
type: "object",
properties: {},
},
},
{
name: "ollama_show",
description: "Show detailed information about a specific model.",
inputSchema: {
type: "object",
properties: {
model: {
type: "string",
description: "The model name to get information about",
},
},
required: ["model"],
},
},
{
name: "ollama_pull",
description: "Pull (download) a model from the Ollama library.",
inputSchema: {
type: "object",
properties: {
model: {
type: "string",
description:
"The model to pull (e.g., 'llama3.2', 'codellama:7b', 'mistral')",
},
},
required: ["model"],
},
},
{
name: "ollama_embeddings",
description: "Generate embeddings for text using an Ollama model.",
inputSchema: {
type: "object",
properties: {
model: {
type: "string",
description:
"The model to use for embeddings (e.g., 'nomic-embed-text', 'mxbai-embed-large')",
},
input: {
type: "string",
description: "The text to generate embeddings for",
},
},
required: ["model", "input"],
},
},
];
export const toolNames = [
"ollama_generate",
"ollama_chat",
"ollama_list",
"ollama_show",
"ollama_pull",
"ollama_embeddings",
] as const;
export type ToolName = (typeof toolNames)[number];
