qURL™ for MCP
Secure access, built into your agent.
Let access be part of the task your agent performs: issue a resource link, use it, and withdraw it when the work is done. qURL brings these controls into your MCP client as tools the agent can call.
How it works
Add one server to your config.
Install Node.js, add this entry to an MCP client that supports local stdio servers, and replace the key placeholder with your API key. Restart the client to load the tools. The second tab shows the arguments for a concrete link-creation call.
{
"mcpServers": {
"qurl": {
"command": "npx",
"args": ["-y", "@layervai/qurl-mcp@0.4.1"],
"env": { "QURL_API_KEY": "lv_live_xxx" }
}
}
}Agent → tool → access
This example uses an existing URL configured for access through LayerV. Creating a link does not start a connector or change that URL’s public exposure. For a local app, start with the CLI quickstart.
The shown MCP release uses resource IDs returned by its URL tools. It does not accept CLI CRIDs in mint_link; use the SDK or API examples to issue links for a CLI-published resource.
Local stdio clients include Claude Desktop, Claude Code, and Cursor. Follow your client’s instructions for storing credentials.
Why it’s different
Access tools your agent can call.
The agent asks for itself
The agent can call create_qurl to issue a link for a configured URL. It can resolve supported access tokens with resolve_qurl, using a credential permitted to perform that operation.
Scope the agent’s credential
Give the MCP server only the API scopes it needs. Access links are bearer credentials with the expiry, use limits, and access policy you set.
Revocation is a tool call
Use revoke_qurl_token for one link and delete_qurl to retire a resource. Handle service errors and any retry required to complete session cleanup.
Give your agents a way in.
Add the tools, configure the resource, and let your agent create access links as part of its work.