> ## Documentation Index
> Fetch the complete documentation index at: https://docs.loqate.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

> Reach is Loqate's agent-native verification tool — a CLI and MCP server that verifies addresses, emails, and phone numbers with deterministic scoring and policy-driven recommendations.

## What is Reach?

Reach is GBG's agent-native identity and address verification product. It verifies addresses, emails, and phone numbers in real time — returning a confidence score, a policy-aware recommendation, and a full audit trail in one command.

Reach is delivered as a single binary (`lqt`) that serves three interfaces from one verification engine:

| Interface              | Best for                                  | How it works                                                   |
| :--------------------- | :---------------------------------------- | :------------------------------------------------------------- |
| **CLI**                | Tool loops, scripts, CI/CD                | Single Go binary. Pipes JSON into agent frameworks.            |
| **MCP Server**         | Claude, Cursor, VS Code, agent frameworks | Stdio mode for local agents, HTTP mode for hosted deployments. |
| **Claude Agent Skill** | Claude users                              | Plain language interface installed via `npx skills add`.       |

All three interfaces share the same scoring engine — same input produces the same confidence score and recommendation regardless of which interface you use.

## How it works

Every verification request follows the same flow:

<Steps>
  <Step title="Submit contact data">
    Send an address, email, phone number — or any combination — via the CLI, MCP tool, or skill.
  </Step>

  <Step title="Verification">
    Reach calls the Loqate APIs to verify each field and compute a confidence score between 0 and 1.
  </Step>

  <Step title="Policy evaluation">
    The confidence score is evaluated against the active policy's thresholds to produce a recommendation: `accept`, `review`, or `reject`.
  </Step>

  <Step title="Structured response">
    A JSON response is returned with per-field scores, an overall recommendation, match-level data, and an audit trail your agent can act on.
  </Step>
</Steps>

## Verification capabilities

<CardGroup cols={2}>
  <Card title="Address Verification" icon="location-dot">
    Confidence scoring, AVC match level, verification status, premise-level matching across 245+ countries.
  </Card>

  <Card title="Email Validation" icon="envelope">
    Disposable domain detection, fraud risk flags, deliverability assessment, catch-all detection.
  </Card>

  <Card title="Phone Validation" icon="phone">
    Number type detection (mobile/landline), carrier identification, E.164 normalization.
  </Card>

  <Card title="Combined Verification" icon="shield-check">
    Verify address, email, and phone in a single request with a unified confidence score and recommendation.
  </Card>
</CardGroup>

## Example output

A single `verify` call returns structured JSON that agents can parse and act on:

```json theme={null}
{
  "address": {
    "confidence": 0.95,
    "recommendation": "accept",
    "avc": "V44-I44-P7-100",
    "verification_status": "V",
    "match_level": "premise"
  },
  "email": {
    "confidence": 0.15,
    "recommendation": "reject",
    "response_code": "Valid",
    "risk": "high",
    "flags": ["disposable_domain"]
  },
  "phone": {
    "confidence": 0.93,
    "recommendation": "accept",
    "verified": true,
    "number_type": "mobile"
  },
  "overall": {
    "recommendation": "reject",
    "confidence": 0.15
  },
  "policy": "strict"
}
```

The `overall.recommendation` is the lowest-confidence field's recommendation — if any field fails the policy threshold, the overall recommendation reflects that.

*This is an illustrative example. See the [lqt documentation](https://github.com/gbgplc/lqt) for full schema and response details.*

## What you need

* A **Loqate API key** — [sign up](https://account.loqate.com) or use an existing key
* The **`lqt` binary** — [installation guide](/ai-agents/cli/installation)
* Optionally, an **Anthropic API key** for the `parse` command (address parsing via Claude Haiku)

The CLI and MCP server are free. Verification calls consume credits from your Loqate account.