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

# Examples

Three prompts you can paste straight into a Claude chat once the LoneScale connector is connected. Each shows the tool Claude is expected to pick and the rough shape of the answer.

## 1. Look up a company

> *"Look up the company at **stripe.com** on LoneScale — I want the name, industry, headcount, and LinkedIn URL."*

**What Claude does:** calls `public-api-companies-search` with `domain: "stripe.com"`.

**Sample response:**

```json theme={null}
{
  "linkedinId": "stripe",
  "linkedinUrl": "https://www.linkedin.com/company/stripe",
  "name": "Stripe",
  "domain": "stripe.com",
  "industry": "Financial Services",
  "staffCount": 8000,
  "headcount": { "engineering": 2400, "sales": 600, ... }
}
```

Add **"and break headcount down by function"** to your prompt and Claude will set `enrich: true`, which returns a function-level breakdown.

## 2. Enrich a list of contacts with email + phone

> \_"Here are 3 LinkedIn URLs:
>
> * [https://www.linkedin.com/in/patrickcollison](https://www.linkedin.com/in/patrickcollison)
> * [https://www.linkedin.com/in/johncollison](https://www.linkedin.com/in/johncollison)
> * [https://www.linkedin.com/in/cristinacordova](https://www.linkedin.com/in/cristinacordova)
>
> Enrich each one with email and phone using LoneScale."\_

**What Claude does:**

1. Calls `public-api-enrich` with `enrichment_type: ["email", "phone"]` and the 3 contacts. Gets back a `lonescale_job_id`.
2. Polls `get-job-result` every few seconds until the status flips from `pending` to `ready`.
3. Shows you a table with the enriched email + phone for each contact.

Enrichment is **waterfall** — LoneScale tries multiple providers and returns the first hit per contact. Most jobs finish in under 60 seconds.

When no provider returns an email, the result falls back to a free **predicted email** built from the contact's name and company domain, flagged with `most_probable_email_status: "predicted"` and accompanied by `predicted_email`, `email_pattern`, and `predicted_email_score`. Predicted emails don't consume credits.

## 3. Source contacts at a company

> *"Find 10 engineering managers at **figma.com** based in the US, and show me their LinkedIn URLs and current titles."*

**What Claude does:**

1. Calls `public-api-contact-sourcing` with `company_domain: "figma.com"`, `personas: ["engineering manager"]`, `included_locations: ["United States"]`, `limit: 10`. Gets back a `lonescale_job_id`.
2. Polls `get-job-result` until the job is `ready` (sourcing typically takes 1–3 minutes).
3. Returns the list of sourced contacts inline.

Chain it with prompt 2 ("…now enrich those 10 contacts with email") and Claude will pipe the LinkedIn URLs from the sourcing result straight into `public-api-enrich`.

## Tips

* **Be explicit about LoneScale.** Mentioning "LoneScale" in the prompt makes Claude pick the right connector when you have several connected.
* **Don't interrupt async jobs.** While Claude is polling `get-job-result`, don't send a new message — it will lose the `lonescale_job_id` and have to restart.
* **Custom metadata.** All tools accept an optional `custom` map that's echoed back in webhook payloads. Useful when you're chaining LoneScale with a CRM update further downstream.