MCP Server
Connect AI agents like Claude Desktop, Cursor, and Windsurf to your TestAssets data using the Model Context Protocol (MCP).
What is MCP?
Model Context Protocol
The Model Context Protocol (MCP) is an open standard that allows AI assistants to connect to external data sources and tools. By connecting your TestAssets account via MCP, AI agents can read your asset lists and search your test data — enabling workflows like generating test plans, reviewing coverage, or finding specific test assets by name or notes.
Prerequisites
- A TestAssets account with at least one asset list.
- An API key — generate one from Settings → API Keys.
- An MCP-compatible client (Claude Desktop, Cursor, Windsurf, etc.).
Available Tools
Nine tools are exposed by the MCP server, covering read, hierarchy, and write operations
list_asset_listsList all asset lists you own or have access to, along with your role (owner, admin, editor, viewer).
get_assetsRetrieve all assets in a specific list. Supports an optional search filter that matches on both name and notes.
search_assetsSearch assets by name or notes across all accessible asset lists. Returns up to 100 results.
get_asset_detailGet a single asset with full context: breadcrumb path (parent chain), direct children, and sibling count — ideal for AI agents that need to understand where an asset sits in the hierarchy.
get_asset_treeReturn an entire asset list (or a subtree starting from a given asset) as a nested JSON hierarchy. Capped at 1,000 items per request.
create_asset_listCreate a new asset list. The API key user becomes the owner. Requires a title and optional description.
create_assetCreate a new asset in a list, optionally nested under a parent. Requires Owner, Admin, or Editor role. Names are limited to 200 characters.
update_assetUpdate an asset's name, comment, or parent. Requires Owner, Admin, or Editor role. Self-parenting is prevented.
delete_assetDelete an asset (and its children, via cascade). Requires Owner or Admin role.
Configuration
Add the following to your MCP client configuration
Claude Desktop
{
"mcpServers": {
"testassets": {
"url": "https://api.testassets.io/v1/mcp-server",
"headers": {
"apikey": "YOUR_API_KEY"
}
}
}
}Cursor
{
"mcpServers": {
"testassets": {
"url": "https://api.testassets.io/v1/mcp-server",
"headers": {
"apikey": "YOUR_API_KEY"
}
}
}
}Windsurf
{
"mcpServers": {
"testassets": {
"serverUrl": "https://api.testassets.io/v1/mcp-server",
"headers": {
"apikey": "YOUR_API_KEY"
}
}
}
}VS Code
{
"servers": {
"testassets": {
"type": "http",
"url": "https://api.testassets.io/v1/mcp-server",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Two options: (1) Provide a static API key via the Authorization header as shown above — fastest setup. (2) Omit the headers block and VS Code will open a browser-based OAuth flow where you paste your API key once; the issued token is reused automatically afterwards.
GitHub Copilot CLI
{
"mcpServers": {
"testassets": {
"type": "http",
"url": "https://api.testassets.io/v1/mcp-server",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Two ways to add the server: (1) Run `copilot mcp add testassets --transport http https://api.testassets.io/v1/mcp-server --header "Authorization: Bearer YOUR_API_KEY"` in your terminal, or (2) inside an interactive Copilot session, type `/mcp add`, choose HTTP, and paste the URL plus the Authorization header. Edit the config file at ~/.copilot/mcp-config.json to make the change permanent.
Replace YOUR_API_KEY with your actual API key from Settings → API Keys.
Endpoint Details
https://api.testassets.io/v1/mcp-server