Skip to main content
LumisReach runs a hosted Model Context Protocol server. Point any MCP client at it with an API key and your agent can place AI phone calls, run AI text conversations, buy numbers, and read call history — no SDK, no local process to run.

Endpoint

Every tool is a thin wrapper over the v1 REST API using your own key, so scopes, rate limits and tenant isolation behave exactly as they do for a direct HTTP call.

Access requirements

1

A LumisReach account

Sign in at app.lumisreach.com. Your workspace is what the agent acts inside — calls, contacts and transcripts land there.
2

An API key

Settings → Developer API (app.lumisreach.com/settings/api). The key is shown once; store it somewhere safe. A key created this way is unrestricted and works on every tool below.
3

A payment method for paid actions

Reading data is free. Placing calls, sending texts and buying numbers draw on your wallet, so add a card under Settings → Billing first. See pricing.
Nothing else is required to use the shared senders: quick calls and quick texts go out from LumisReach-owned numbers, so an agent can make its first call minutes after the key is created.

Connect a client

Tools

Quick Agent Actions

These need no phone number of your own — they run from shared LumisReach senders. This is what an agent should reach for by default.

Primitives

For callers that already own a number and want to drive it directly.

Example

“Call +1 415 555 0123 and find out whether they still want the quote. Try twice if nobody answers.”
The agent calls start_ai_phone_call with the goal and retry_attempts: 1, gets back a job id, and polls get_quick_action_status until the target reports completed with a summary of what was said.

Troubleshooting