Web product
Aggors: a merchant evidence API for AI agents
Aggors collects 14 observable signals about an online store and hands them to your agent as evidence - with unknowns labelled as unknown, and no safety verdict attached.
The problem it solves
Agents are increasingly asked to act on a merchant: place an order, follow a link, recommend a store, approve a payout. To do that responsibly they need something to reason over, and what most services hand them is a score - a single number that has already collapsed a dozen unrelated measurements and a pile of missing data into one confident-looking verdict.
The trouble with a score is that it hides its own uncertainty. It cannot tell you whether the store looked fine or whether the check simply failed. It bakes in a risk tolerance that belongs to whoever built the scorer, not to whoever is about to act. And it gives an agent no way to explain its decision afterwards.
What Aggors does
Aggors is deliberately not a scorer. It is read-only evidence infrastructure: it observes what can be observed about a store, reports it with sources and timestamps, and stops there. Interpretation and policy stay with your application.
The service exposes 15 MCP tools at https://aggors.com/mcp. The primary tool, check_merchant, returns all 14 signals in a single call; individual named tools are available when an agent only needs one observation.
The 14 observation areas
- UCP profile
- HTTPS connectivity
- DNS blocklists
- Website content
- Domain patterns
- Domain age
- Archive history
- DNS records
- HTTP security headers
- Registration quality
- Hosting reputation
- Merchant verification
- Policy pages
- TLS certificate details
Missing evidence is never negative evidence
This is the design commitment the whole product is built around. Every signal comes back explicitly marked as observed, unavailable, or unknown, alongside freshness metadata describing cache status and whether the measurement could be taken at all. A check that could not run reads as a check that could not run - not as a red flag. An agent consuming Aggors can therefore distinguish "this store has no policy pages" from "we could not reach the store", which a score cannot.
Who it is for
Developers building agentic commerce flows, marketplace and payout risk tooling, browser or shopping assistants, and anyone who has to document why an automated decision was made. If your compliance story requires showing your work, evidence with timestamps and sources is a better substrate than a number.
There is also a free consumer-facing checker on aggors.com for anyone evaluating a single store by hand before buying.
Pricing
Metered by MCP unit - one unit per successful call. Failed calls do not consume units, and results are reusable from cache for up to 24 hours.
- Starter - $29/month, 1,000 units, then $0.03 each
- Growth - $99/month, 5,000 units, then $0.02 each
- Scale - $299/month, 20,000 units, then $0.015 each
Authentication is by API key or OAuth 2.1 bearer token with the merchant:read scope.
Common questions
Will Aggors tell me whether a store is safe?
No, and that is on purpose. Aggors does not issue merchant safety verdicts. It reports what was observed, what was not, and when. Context and risk tolerance belong to you.
How does an agent connect to it?
Over MCP, at https://aggors.com/mcp, authenticated with an API key or an OAuth 2.1 bearer token carrying the merchant:read scope. Any MCP-compatible client can call the tools directly.
Do I get charged for checks that fail?
No. A unit is consumed only on a successful call.
How fresh are the results?
Every response carries freshness metadata, and cached results are reused for at most 24 hours. You always know how old a measurement is rather than having to assume.
Is there a free tier?
Not for the API. The consumer checker on aggors.com is free to use for one-off manual lookups.