sv-number/skills

Give your AI agent a phone number: order a private number in 200+ countries over the API, read the SMS verification code, hand the number back. The widest country coverage in the category, checkable with one API call.

SV Number agent skill – what it is

A tiny, self‑contained skill that lets an autonomous AI agent obtain a disposable phone number, receive the SMS verification code for a service, and (optionally) generate a TOTP code – all through a commercial API. The skill follows the Agent Skills specification, so any runner that can read a SKILL.md file (Claude Code, Codex, etc.) can use it without extra SDKs.

Core capabilities

Feature How it works (as described in the README)
Get a number in the right country The agent calls the API endpoint action=getCountryAndOperators to list the 200+ supported countries, then requests a number for the desired country (e.g., Indonesia).
Receive the verification code directly The API returns the code already extracted from the SMS, so the agent never has to parse an inbox or write regexes.
Pay‑per‑code Money is held when the number is ordered and only charged when the code arrives (or released after ~20 min if no code).
Private, one‑time numbers Each activation is tied to a single service request; no other user can see the code.
Dynamic pool awareness The catalogue call also reports the success‑rate and live‑pool size per country, letting the agent avoid low‑yield regions.
Finish 2‑FA with TOTP If the service asks for an authenticator, the agent stores the TOTP secret locally and computes the 6‑digit code (RFC 6238) without sending the secret anywhere.

Typical workflow (as illustrated in the README)

1. getNumber   → ACCESS_NUMBER:id:phone   (order a number, money held)
2. getStatus   → STATUS_WAIT_CODE       (poll every 3‑5 s)
3.               → STATUS_OK:123456      (code received)
4. setStatus 6 → ACCESS_ACTIVATION      (code used, charge stays)
5. setStatus 8 → ACCESS_CANCEL          (no code, money refunded)

The repository ships two ready‑to‑run scripts (order_number.py and order_number.mjs) that implement this loop.

Getting started

# Grab the skill definition
curl -O https://raw.githubusercontent.com/sv-number/skills/main/sv-number/SKILL.md

# Export your API key (obtain it after signing up at sms‑verification‑number.com)
export SVN_API_KEY=your_key_here

Place the sv-number directory where your agent runner expects skills, and invoke the skill from your LLM‑driven workflow.

What it is not

  • It does not provide a full‑duplex phone line – numbers cannot receive calls or hold ongoing conversations.
  • It is not intended for high‑risk use cases such as banking, payment, or government account verification.
  • The underlying service is commercial; the skill itself is MIT‑licensed, but you must fund the API account.

Limitations & constraints

  • You need a paid account on sms-verification-number.com; there is no free tier.
  • Each activation must be closed (setStatus 6 or 8) otherwise the held funds remain locked.
  • Success depends on the provider’s pool health – the skill helps you avoid bad pools, but delivery is never 100 % guaranteed.

Quick example (from the README)

> Read the sv-number skill and register me a Telegram account on an Indonesian number.

ordered +62 838 1234 5678
code 123456
account created

The agent asks the skill for an Indonesian number, receives the code 123456, and can complete the Telegram signup automatically.

Links

Related

  • Project
    nullpointexception-i/agent-sphere

    AgentSphere is a Java‑Spring/React platform that runs LLM‑driven ReAct loops, supports multi‑agent sub‑runs, real browser automation via a Chrome extension, model routing, persistent multi‑level memory, human‑in‑the‑loop clarification, OIDC SSO, and an embeddable chat widget.

  • Project
    oleksiijko/pmb

    PMB is an open‑source Python library that gives LLM‑based coding assistants a local, SQLite‑backed memory. It automatically indexes code, PDFs, git history, and user‑recorded facts, then injects the most relevant pieces back into the agent via the Model‑Context‑Protocol. The system works offline, offers a hybrid BM25‑+‑vector search (≈35 ms warm latency), provides a visual dashboard, and includes hooks that both recall memory before the model thinks and auto‑write observed actions. Install with `pip install pmb-ai`, run `pmb setup` to wire an agent, and the agent will remember decisions, lessons, and project context across restarts without any API keys or cloud calls.

  • Project
    jinzijian/EvoTrace

    EvoTrace is a local‑first tool that imports Claude Code or Codex session histories, mines coherent coding episodes, and compiles them into verified, Docker‑ready tasks, preference data, and reward candidates for AI training and evaluation. It runs on the DeepSeek Harness UI, uses a four‑agent review pipeline, and keeps all data under the user’s control.