Trieve MCP Server

1# trieve-mcp-server
2 
3[![smithery badge](https://smithery.ai/badge/trieve-mcp-server)](https://smithery.ai/server/trieve-mcp-server)
4 
5A Model Context Protocol (MCP) server that provides agentic tools for interacting with the Trieve API. This server enables AI agents to search and interact with Trieve datasets through a standardized interface.
6 
7## Features
8 
9- Search across Trieve datasets using semantic search
10- List and access dataset information
11- Support for both environment variables and command-line arguments
12- Built with TypeScript for type safety and better developer experience
13 
14## Installation
15 
16### Installing via Smithery
17 
18To install Trieve for Claude Desktop automatically via [Smithery](https://smithery.ai/server/trieve-mcp-server):
19 
20```bash
21npx -y @smithery/cli install trieve-mcp-server --client claude
22```
23 
24### Manual Installation
25```bash
26npm install trieve-mcp-server
27```
28 
29## Configuration
30 
31Copy the `.env.dist` file to `.env` and fill in your Trieve credentials:
32 
33```bash
34cp .env.dist .env
35```
36 
37Required environment variables:
38- `TRIEVE_API_KEY`: Your Trieve API key from dashboard.trieve.ai
39- `TRIEVE_ORGANIZATION_ID`: Your Trieve organization ID from dashboard.trieve.ai
40 
41Optional environment variables:
42- `TRIEVE_DATASET_ID`: Specific dataset ID to use (if not provided via CLI)
43 
44Command-line arguments (override environment variables):
45```bash
46trieve-mcp-server --api-key <your-api-key> --org-id <your-org-id> [--dataset-id <dataset-id>]
47```
48 
49## Usage
50 
51### Starting the Server
52 
53```bash
54trieve-mcp-server
55```
56 
57### Available Tools
58 
59#### search
60Search through a specified Trieve dataset.
61 
62Parameters:
63- `query` (string): The search query
64- `datasetId` (string): ID of the dataset to search in
65- `searchType` (string, optional): "semantic" (default), "fulltext", "hybrid", or "bm25"
66- `filters` (object, optional): Advanced filtering options
67- `highlightOptions` (object, optional): Customize result highlighting
68- `page` (number, optional): Page number, default 1
69- `pageSize` (number, optional): Results per page, default 10
70 
71Example:
72```json
73{
74  "query": "example search query",
75  "datasetId": "your-dataset-id",
76  "searchType": "semantic",
77  "page": 1,
78  "pageSize": 10
79}
80```
81 
82### Available Resources
83 
84The server exposes Trieve datasets as resources with the following URI format:
85- `trieve://datasets/{dataset-id}`
86 
87## Usage with Claude Desktop
88 
89The Trieve MCP Server supports MCP integration with [Claude Desktop](https://modelcontextprotocol.io/quickstart/user). Place the following in your Claude Desktop's `claude_desktop_config.json`.
90 
91```json
92{
93  "mcpServers": {
94    "trieve-mcp-server": {
95      "command": "npx",
96      "args": ["trieve-mcp-server@latest"],
97      "env": {
98        "TRIEVE_API_KEY": "$TRIEVE_API_KEY",
99        "TRIEVE_ORGANIZATION_ID": "$TRIEVE_ORGANIZATION_ID",
100        "TRIEVE_DATASET_ID": "$TRIEVE_DATASET_ID"
101      }
102    }
103  }
104}
105```
106 
107Note: Instead of environment variables, `--api-key`, `--org-id`, and `--dataset-id` can be used as command-line arguments.
108 
109Once Claude Desktop starts, attachments will be available that correspond to the [datasets available to the Trieve organization](https://docs.trieve.ai/guides/create-organizations-and-dataset). These can be used to select a dataset. After that, begin chatting with Claude and ask for information about the dataset. Claude will use search as needed in order to filter and break down queries, and may make multiple queries depending on your task.
110 
111## Development
112 
113### Setup
114 
1151. Clone the repository
1162. Install dependencies:
117```bash
118npm install
119```
1203. Copy `.env.dist` to `.env` and configure your credentials
1214. Build the project:
122```bash
123npm run build
124```
125 
126### Scripts
127 
128- `npm run build`: Build the TypeScript project
129- `npm run watch`: Watch for changes and rebuild
130- `npm run test`: Run tests
131- `npm run inspector`: Run the MCP inspector for debugging
132 
133## License
134 
135MIT
136