Realask - verified answers from the physical world
RemoteAbout Realask - verified answers from the physical world
Realask is a paid MCP server and HTTP API for AI agents to verify current facts or arrange requests by calling US local businesses. It returns structured results with evidence, timestamps, confidence scores, and details that could not be verified, rather than a call transcript.
Key features
- 01Calls US local businesses
- 02Returns structured verification results
- 03Includes evidence, timestamps, and confidence
- 04Free quotes and per-job payment
- 05Can arrange or reserve requests
Use cases
- 01Checking current stock and prices
- 02Verifying same-day availability or wait times
- 03Asking about unpublished capabilities or policy exceptions
- 04Requesting a booking, hold, or cancellation
MCP support
Realask is a remote MCP server accessed over streamable HTTP. Its tools support quoting, purchasing, job-status checks, and result retrieval; purchases use x402 payments.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://realask.net/mcpclaude mcp add --transport http danelas-realask https://realask.net/mcp{"mcpServers":{"danelas-realask":{"type":"http","url":"https://realask.net/mcp"}}}Open in Cursor{"servers":{"danelas-realask":{"type":"http","url":"https://realask.net/mcp"}}}Open in VS Code{"mcpServers":{"danelas-realask":{"type":"http","url":"https://realask.net/mcp"}}}{"mcpServers":{"danelas-realask":{"serverUrl":"https://realask.net/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
5| Parameter | Type | Required | Description |
|---|---|---|---|
| limits | object | No | |
| limits.max_businesses | integer | No | |
| limits.deadline_seconds | integer | No | |
| limits.max_call_attempts_per_target | integer | No | |
| request | string | Yes | Natural-language question, e.g. 'Does ABC Tire in 33312 have two Michelin Pilot Sport 4S 235/40R19 in stock today and what is the installed price?' |
| location | object | Yes | |
| location.lat | number | No | |
| location.lng | number | No | |
| location.city | string | No | |
| location.region | string | No | |
| location.country | string | No | |
| location.postal_code | string | No | |
| location.radius_miles | number | No | |
| callback_url | string | No | |
| verification_mode | enum | No | auto reuses fresh cached facts when safe; live_verified forces a new calldigital_fast | live_verified | auto |
Details
From the project READMERealask MCP
Ask a US local business a current question by phone and get a typed, timestamped, confidence-scored answer back — not a transcript.
Realask is a paid MCP server and HTTP API for AI agents. When a fact changes faster than the web reflects it, or was never published at all — today's stock, a real all-in price, same-day availability, an edge-case capability, a policy exception, the current wait — an agent asks Realask, Realask calls the business, and the agent gets back structured facts with evidence, a timestamp, a confidence score, and an honest list of what could not be verified.
This repository is the public front door: the MCP manifest, connection details, and docs. The service itself is hosted at https://realask.net and is closed-source.
- Website: https://realask.net
- Machine docs: https://realask.net/llms.txt · https://realask.net/v1/capabilities · https://realask.net/openapi.json
- Payment: https://realask.net/.well-known/x402
- MCP registry:
io.github.danelas/realask
Connect
Realask is a remote MCP server over streamable HTTP. Nothing to install.
{
"mcpServers": {
"realask": {
"type": "streamable-http",
"url": "https://realask.net/mcp"
}
}
}
Quoting is free and unauthenticated. Buying a quote requires payment (see below), so the purchase tool returns an x402 challenge to any agent that cannot pay for itself.
Tools
| Tool | What it does |
|---|---|
quote_real_world_answer | Free, immutable price quote for verifying a current fact about a US local business. |
quote_real_world_action | Free price to arrange something — book, hold, request a quote, cancel. Requires explicit authorization, a customer name and callback number, and a price ceiling. |
purchase_quote | The single purchase path for either kind of quote. Returns job_id + job_token. |
get_verification_status | Job progress. Poll every 15–30s; live calls take minutes. |
get_verification_result | The final answer: typed facts, evidence excerpts, timestamps, confidence, limitations. |
Paying
Pricing is per job and quoted up front. Payment is x402 —
exact scheme, USDC on Base (eip155:8453) — so a wallet-enabled agent pays inline:
POST /v1/quotes -> free: price, scope, expiry
POST /v1/jobs -> 402 with payment requirements -> sign -> 202 { job_id, job_token }
GET /v1/jobs/{id}/result
Cached, recently verified answers are offered at a lower price; live_verified forces a
fresh call. Full requirements: https://realask.net/.well-known/x402
What it will not do
Realask arranges and reserves; it never buys and never bargains.
- Never negotiates, haggles, or asks for a discount
- Never pays, and never gives card or bank details — the customer pays the business directly
- Never agrees above the maximum price you set
- Never substitutes a different item, time, or business than the one you specified
- Shares only the customer name and callback number you supply, nothing else
- Commits to exactly one business per job, and can never commit twice
No mass calling, no marketing calls, no call recordings.
For businesses
If we called you and you would rather we did not, press 1 during the call or use https://realask.net/opt-out. Businesses can also connect a live data feed so agents are answered from your own data, with you as the source, and we stop calling: https://realask.net/merchants
Questions: [email protected] · https://realask.net/privacy · https://realask.net/terms
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.