Skip to content
Rate AI Agent
Open menu

Legal

Terms of service

The rules for using Rate AI Agent, listing an agent and requesting a paid evaluation. Last updated [date].

Draft: not yet reviewed by a lawyer

This page describes what the site actually does today, but it is not legal advice and has not been checked for the laws that apply to us or to you. Text in [square brackets] is a placeholder we still have to fill in. It must be reviewed before we collect emails or take payments.

1. Who we are

Rate AI Agent (rateaiagent.com, “the site”, “we”) is run by [Company legal name], [registered address]. You can reach us at [Contact email]. These terms are governed by the laws of [jurisdiction].

By using the site you agree to these terms. If you don't agree, please don't use it.

2. What the site does

  • It is a free directory of AI coding agents, with independent evaluation results where we have run them.
  • Every figure is labelled with where it came from: an independent evaluation, the publisher, the community, or not evaluated. The methodology explains how results are produced and their limits.
  • Results describe how one version of an agent, with one model, performed on our tasks under the recorded conditions. They are information, not a guarantee of how an agent will work for you, and not a recommendation to buy anything.

3. Accounts

  • You sign in with a one-time link sent to your email. You need an account to list an agent or request an evaluation.
  • Keep access to your email account secure. Anything done after signing in with your link is treated as done by you.
  • We may suspend accounts used for spam, abuse or attempts to interfere with evaluations.

4. Listing an agent (free)

  • Listing is free. Every submission is reviewed before it appears, and starts as pending review.
  • You must be allowed to submit the agent (for example, you build it or work for its publisher), and the details must be accurate and not misleading.
  • Approved details are shown labelled publisher-provided, with a link to your public pages as the source. They are never shown as measured results.
  • Reviewers may edit details to match your public pages or our format, or decline a submission. Reviewers never review agents from their own organisation.
  • You give us a non-exclusive, worldwide, royalty-free licence to publish the details you submit (name, description, links, logos and similar) on the site and in its public API. You keep ownership of them.
  • A listing never includes a score. Scores come only from an independent evaluation.

5. Paid evaluations

How it works

  • Requesting a quote on /evaluate is free and commits you to nothing.
  • If we can run your agent, we email you a fixed price and a payment link. You pay only if you accept the quote. Payments are processed by Stripe. We never see or store your card details.
  • We aim to run and publish within [7] days of payment. This is a target, not a guarantee.
  • You must have the right to let us run the agent version you name, and the instructions you give us must not ask us to do anything unlawful.

What payment does and does not buy

  • Payment buys one evaluation run: one agent version with one model, on the current suite, under our standard conditions.
  • The result is always published, whatever it shows. You agree to this when you request a quote.
  • Payment never changes the score, ranking, levels, badges or review. Paid runs use the same tasks, sandbox, limits and grading as every other run.
  • Every paid result shows who funded it (“Funded by: Publisher”).
  • Hidden tests and task repositories are never shared, including with paying publishers. You receive the diff, transcript and test output for each of your agent's attempts.

Errors, disputes and removal

  • If a result is wrong because of our harness, we fix it and re-run at our cost before publishing.
  • If you disagree with a result, tell us at [Contact email]. If the cause is in our harness, we re-run for free. Otherwise the result stands.
  • A published result can't be removed or hidden. You can request an evaluation of a newer version, and the profile shows the latest run first with the full history.

Refunds

[Refund policy to be decided. For example: a full refund if we can't run the agent after payment; no refund once a run has started.] A refund never removes a result that has been published.

The full rules are in our conflict-of-interest policy.

6. Using the site and its data

  • You may read, share and quote results, with a link back to the agent's page so the conditions stay attached.
  • The JSON API is a preview. It may change or be rate-limited. Keep the provenance and illustrative fields when you republish its data.
  • Don't try to obtain the hidden tests or task repositories, game an evaluation, overload the site, or get around its access controls.
  • Agents marked Example are fictional and their results are simulated to show the format.

7. Disclaimers and liability

The site and its results are provided “as is”. We work hard to make results accurate and reproducible, but we can't promise the site will be error-free or always available. [Limitation of liability wording to be added after legal review, including any cap on liability for paid evaluations.]

8. Changes to these terms

We may update these terms. When we do, we'll change the date at the top of this page. [Decide how you'll tell users about significant changes, for example by email.]

9. Contact

Questions about these terms: [Contact email]. How we handle personal data is in the privacy policy.