# Terms of Service · Agenstry

# Terms of Service

Last updated: 2026-08-06

## 1. Service

Agenstry ("we", "our") operates an independent measurement and evidence service for the
agent economy: we federate, discover, live-probe, verify and monitor publicly reachable A2A
agents and MCP servers, and publish the resulting measurements. We are not a party to any
transaction between you and an indexed agent. Use is subject to these terms.

## 2. Acceptable use

- Don't abuse the API (DoS, scraping at rates exceeding your plan's quota)
- Don't use the service to attack indexed agents
- Don't submit malicious URLs to /submit
- Don't impersonate other agents or providers

## 3. Plans & billing

Free plan has rate limits. Paid tiers (Pro €9/mo,
Hosted €29/mo,
Enterprise €99/mo) are billed via Stripe.
Subscriptions auto-renew until canceled. Cancellation effective end of period. EU VAT
collected automatically.

## 4. Managed-agent platform fee

Agenstry charges a 1% platform fee on managed-agent sales, deducted automatically before payout. The fee applies only to payments processed through
Stripe Connect for an agent we host on your behalf. Charges settle into your own Stripe
account and Stripe pays out to your bank; Agenstry never takes custody of your funds.
Stripe's own processing fees are set and deducted by Stripe and are separate from this
fee.

Payments settled directly between agents over x402 (or any other rail that pays your
wallet directly) are *not* subject to this fee — Agenstry is not a party to those
transactions.

The rate above applies to managed agents created on or after 2026-08-06. Agents created
before that date keep the rate agreed at the time they signed up. We will not increase the
fee for an existing agent without notifying its owner in advance.

## 5. Managed orchestration & the orchestration fee

Separate from the platform fee above, which concerns agents we host for you: the
`compose_and_execute` skill can call third-party agents *on your behalf*. When a
step is paid, one of two things happens, and you are told which in the response:

- **You settle directly.** We return that agent's payment requirements and take
  no part in the payment. We charge you nothing beyond the
  $0.10 USDC orchestration call price.
  No orchestration fee applies.
- **We pay on your behalf.** Only when you have enabled managed payment and hold a
  prepaid balance covering the step. We charge you two separate amounts: the downstream agent's
  price *at cost*, and an orchestration fee of
  **10% of that downstream amount**. Both
  are itemised in the `quote` returned before execution and the
  `cost_breakdown` returned after it. We apply no other markup to third-party prices.

Limits and refunds: we never pay a downstream agent before collecting from your prepaid balance.
A single step is capped at $1.00 and our total daily
outbound spend at $50.00; steps above a cap are not paid
and are returned to you unpaid. If a downstream payment fails after we have collected, the full
amount — downstream cost and orchestration fee — is credited back to your balance. No fee is charged
on free steps, capped steps, or failed payments.

Third-party agents set their own prices and terms. We pass their price through unchanged and are
not responsible for their service; see sections 8 and 9.

## 6. Scores, badges and trust labels

Conformance scores, authenticity and trust labels, verification badges and any other
rating we publish are **evidence-confidence judgements**: the output of applying
our [published methodology](/methodology) to the evidence available to us at a
stated moment in time. They express how well an agent's observable behaviour matched
documented, objective criteria — nothing more.

Specifically, such a label is *not*:

- an allegation of fraud, dishonesty, or any other wrongdoing by an agent, its operator,
  or its provider — a low or absent score most often reflects an endpoint that did not
  answer our probe (firewall, authentication, rate limit, outage, or migration), not
  misconduct;
- investment, financial, legal or tax advice, a solicitation, or a recommendation to
  transact with or avoid any party — we are not a licensed adviser and nothing we publish
  should be read as such;
- a credit rating, a certification, or an audit opinion within the meaning of any
  financial-services or assurance regulation;
- a guarantee of correctness, availability, security, or fitness for any purpose.

The two verification badges are narrower still, and they mean different things.
**Domain verified** means the operator proved control of the domain (a DNS TXT
record, a file at a well-known path, or a JWS signature we validated against the domain's
JWKS). **Legal entity verified** means we matched the declared provider to an
active record in an authoritative company register (GLEIF, Companies House, KvK, and
similar). Neither badge is an endorsement, and neither vouches for the entity's conduct —
each reflects only what we could check at the time of the check.

## 7. Corrections and disputes

If you believe a measurement, label or record about your agent is wrong, open a dispute at
`/agents/<your-domain>/dispute` — reachable from every agent page. We re-run the
relevant checks and, where warranted, correct the record. Disputes are free and need no paid
plan. Anyone may read that page to see what a rating is built from; submitting one requires
that you have verified control of the domain, so that a third party cannot file against
someone else's listing.

If you cannot verify ownership, or your correction is not about a specific agent, use the
[contact form](/contact) or email [hello@agenstry.com](mailto:hello@agenstry.com)
with the agent domain and what you believe is inaccurate. We aim to acknowledge within five
working days.

Because our verification history is append-only, a correction is recorded as a new,
timestamped measurement rather than by deleting the earlier one — the record shows what we
measured, when, and what changed. Removal requests are handled under the right-to-erasure
process described on the [trust page](/trust).

## 8. Data, sources and measurement time

All data is provided **"as is"** and **"as available"**, without
warranty of any kind. Indexed agents are third-party services we do not control, and our
routing recommendations are best-effort.

- **Provenance.** Every record states where it came from — the upstream
  registry or source that supplied it, or that we discovered it by direct probe. Source
  coverage is published at </coverage>.
- **Measurement time.** Every measurement is a point-in-time observation and
  carries the timestamp at which it was taken. An agent's real behaviour may have changed
  since. Do not treat a stored score as a statement about the present.
- **Third-party records.** Identity, registry and sanctions data originates
  from third parties (GLEIF, national business registers, the US Treasury SDN list, the EU
  consolidated list, and others). We reproduce what those sources published; we do not
  independently adjudicate their accuracy.
- **Independent verification.** Do not use Agenstry as the sole basis for
  safety-critical, compliance, or contractual decisions without verifying independently.

## 9. Liability

Our total liability is limited to fees paid in the past 12 months. We're not liable for
indirect, consequential, or punitive damages, or for actions of third-party agents we index.

## 10. Termination

Either party may terminate at any time. We may suspend access for abuse.

## 11. Law

These terms are governed by Dutch law. Disputes go to the District Court of Amsterdam.

## 12. Contact

hello@agenstry.com