# discava/mcp-server [Health: Active]

**Category:** 🗺️ Location Services  
**Repository:** https://github.com/discava/mcp-server  
**GitHub Stars:** 1  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/discava-mcp-server

## Description
discava - The Business Directory for AI - mcp server

## Tools
Capabilities this server exposes over MCP:

- **search_businesses** — Search for businesses by query, city, country
- **get_business** — Get full details for a business by ID
- **send_feedback** — Report data quality issues
- **get_rankings** — Top businesses by demand score
- **suggest** — Autocomplete for cities and categories

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "mcp-server": {
    "command": "npx",
    "args": ["tsx","/path/to/discava/mcp/server.ts"]
  }
}
```

## Documentation

## What the discava/mcp-server MCP server does

The discava/mcp-server MCP server connects AI agents to Discava, a business directory for local business discovery. It lets an agent search businesses using a query together with city and country filters, then retrieve a full business record by its ID. The available operations are focused on structured directory access rather than general browsing.

The server also exposes rankings based on demand score, autocomplete suggestions for cities and categories, and a feedback operation for reporting data-quality problems. These capabilities support workflows such as finding businesses in a particular location, narrowing a category, inspecting a selected listing, and flagging inaccurate information.

## How it works

The server runs as a local TypeScript entrypoint and communicates with the Discava API. MCP-compatible clients invoke the exposed tools, while the server forwards requests to the configured API base URL. The default API URL is `https://discava.ai/api/v1`.

A typical interaction can begin with `suggest` to resolve a city or category, continue with `search_businesses` using the resulting context, and use `get_business` to retrieve complete information for a selected business. An agent can call `get_rankings` when it needs businesses ordered by demand score, or `send_feedback` when it identifies a data-quality issue.

## Setup and configuration

The README shows a Claude Desktop configuration that runs the server with `npx`, `tsx`, and a local path to `server.ts`. The same local entrypoint can be registered with Claude Code using its MCP add command. Because the examples contain a placeholder filesystem path, the path must be replaced with the actual location of the repository before use.

Set `DISCAVA_API_URL` to override the API base URL. If it is not set, the server uses `https://discava.ai/api/v1`. The provided setup material does not specify an API key, OAuth flow, or other credential requirement.

## Tools and capabilities

The discava/mcp-server MCP server provides these tools:

- `search_businesses`: searches for businesses by query, city, and country.
- `get_business`: retrieves full details for a business identified by ID.
- `send_feedback`: reports issues with business data.
- `get_rankings`: returns top businesses by demand score.
- `suggest`: provides autocomplete suggestions for cities and categories.

Together, these tools cover discovery, record retrieval, ranking, input assistance, and feedback submission within the Discava directory.

## Limitations and notes

The supplied material does not describe the fields returned by each tool, search ranking behavior beyond the demand-score ranking tool, rate limits, error handling, or authentication requirements. It also does not provide a published package name or a standalone install command. The documented setup therefore relies on a local checkout and a real path to the TypeScript server file. Claude Desktop is explicitly documented; compatibility with other clients is not established by the provided README.

_Full upstream README: https://allmcps.com/mcp/discava-mcp-server/readme_