Temporary phone number and OTP session infrastructure built for AI agents, with structured OTP parsing and native MCP support.
{
"score": 94,
"tier": "Agent Native",
"history": [
{
"date": "2026-03-30",
"score": 94,
"tier": "Agent Native",
"summary": "Certification audit revised upward after confirming a full public error taxonomy, typed SDK exceptions, explicit retry semantics, and per-endpoint rate limit documentation. AgentSIM now clears the Agent Native threshold, with a public status page remaining the main visible gap."
}
],
"breakdown": [
{
"index": "01",
"criterion": "Machine Authentication",
"score": 10,
"max": 10,
"note": "API key auth (asm_live_...) with no OAuth flow or human login required. Agent accounts and environment variable setup are documented as first-class primitives."
},
{
"index": "02",
"criterion": "Autonomous Onboarding",
"score": 10,
"max": 10,
"note": "Free tier requires no credit card, SDK install is a single command, and the quickstart promises a first OTP in under five minutes without support intervention."
},
{
"index": "03",
"criterion": "Structured I/O",
"score": 9,
"max": 10,
"note": "Responses are consistently structured JSON, including parsed OTP payloads with a confidence signal that helps agents decide how to proceed downstream."
},
{
"index": "04",
"criterion": "MCP Support",
"score": 10,
"max": 10,
"note": "MCP support ships on day one with clearly scoped tools for provisioning numbers, waiting for OTPs, and releasing sessions across major agent IDEs."
},
{
"index": "05",
"criterion": "Error Legibility",
"score": 10,
"max": 10,
"note": "Public docs define a consistent error shape, seven named error codes with HTTP status mappings, retry guidance per failure mode, typed SDK exceptions, and Retry-After header support."
},
{
"index": "06",
"criterion": "Agent-Specific Documentation",
"score": 9,
"max": 10,
"note": "The product is framed around zero human assumptions, with examples and quickstart paths written explicitly for Claude and other AI agent workflows."
},
{
"index": "07",
"criterion": "Pricing Transparency",
"score": 9,
"max": 10,
"note": "$0.99 per session, ten free sessions per month, and no monthly fee are clearly stated, with only enterprise pricing moving to contact-sales territory."
},
{
"index": "08",
"criterion": "Rate Limits & Reliability",
"score": 9,
"max": 10,
"note": "Per-endpoint rate limits, response headers, and concurrent session limits are documented, alongside parallel provisioning, webhooks, polling, and number auto-rotation. The remaining gap is a public status or uptime page."
}
]
}null
Certification audit revised upward after confirming a full public error taxonomy, typed SDK exceptions, explicit retry semantics, and per-endpoint rate limit documentation. AgentSIM now clears the Agent Native threshold, with a public status page remaining the main visible gap.
API key auth (asm_live_...) with no OAuth flow or human login required. Agent accounts and environment variable setup are documented as first-class primitives.
Free tier requires no credit card, SDK install is a single command, and the quickstart promises a first OTP in under five minutes without support intervention.
Responses are consistently structured JSON, including parsed OTP payloads with a confidence signal that helps agents decide how to proceed downstream.
MCP support ships on day one with clearly scoped tools for provisioning numbers, waiting for OTPs, and releasing sessions across major agent IDEs.
Public docs define a consistent error shape, seven named error codes with HTTP status mappings, retry guidance per failure mode, typed SDK exceptions, and Retry-After header support.
The product is framed around zero human assumptions, with examples and quickstart paths written explicitly for Claude and other AI agent workflows.
$0.99 per session, ten free sessions per month, and no monthly fee are clearly stated, with only enterprise pricing moving to contact-sales territory.
Per-endpoint rate limits, response headers, and concurrent session limits are documented, alongside parallel provisioning, webhooks, polling, and number auto-rotation. The remaining gap is a public status or uptime page.