Skip to main content
PreviewFor agent & MCP-server developers

Connect agents to private MCP tools.

Let MCP-compatible agents create, resolve, list, and manage qURLs without custom integration code.

The usual way to do this is leaky

Reaching a private resource from MCP usually means one of these workarounds — each of which leaves something standing that shouldn’t be.

  • Publishing an MCP server on a public URL behind a static token
  • Long-lived API keys baked into agent configs
  • Tunnels left running so an agent can reach a local tool

Verify the requester, then open one private route

Requester

A MCP user or agent asks for access to a private resource.

LayerV

Identity and policy are verified before anything becomes reachable.

Private resource

Connected to LayerV with no public inbound path of its own.

Temporary route

A single-use, identity-bound qURL opens for the approved requester, then expires.

Real code you can run today

curl -X POST https://api.layerv.ai/v1/resolve \
  -H "Authorization: Bearer $LAYERV_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "access_token": "at_k8xqp9h2sj9lx7r4" }'

Full, open-source examples live on GitHub →

From zero to first access

  1. 1

    Get an API key

    Create a key with the qurl:write and qurl:resolve scopes.

  2. 2

    Connect the private service

    Register the target the agent needs to reach.

  3. 3

    Mint a qURL per call

    Create a single-use token scoped to the request.

  4. 4

    Resolve headlessly

    The agent POSTs the token to /v1/resolve and gets a target URL it can reach for a short grant.

Available now, and what isn’t yet

Works today

  • Headless agent resolution via POST /v1/resolve
  • AI-agent access policy (allow/deny by agent category)
  • Open-source integration examples on GitHub

Not yet

  • A 1-click MCP install — today this is a build-your-own reference, not a packaged server