Terms of Service

Last updated: August 19, 2026. These terms govern use of the Banana Peel site, console, and API at bananapeel.com, provided by [Banana Peel legal entity]("Banana Peel", "we"). By creating an account or calling the API you agree to them. Bracketed items are pending legal review and will be replaced before these terms are final.

1. The service

Banana Peel is a routing layer for browser automation: you submit a task, and we dispatch it to one or more execution runners — vendor-hosted browser and agent services, extraction APIs, or open-source engines on our compute — selecting and falling back between them. Results, artifacts, and routing telemetry are returned per the API documentation. The docs are part of how the service is described, including retention, rate limits, and credential custody.

2. Accounts and API keys

  • You are responsible for activity under your account and API keys. Keys are shown once and stored hashed; keep them secret, and revoke any key you suspect is exposed (Console → API Keys).
  • Agent-created accounts (via agent signup) start with welcome credits and must be claimed by a human to keep running once those credits are exhausted or the claim window lapses.
  • You must provide accurate billing information to purchase credits.

3. Acceptable use

  • No unlawful automation. Do not use the service to break the law, to access systems or data you are not authorized to access, to commit fraud, or to harass, defraud, or impersonate others.
  • Third-party sites are your responsibility.Agents act on third-party websites at your instruction. You are responsible for ensuring your automation of a target site is permitted — by that site's terms, by contract, or by law that applies to you. Banana Peel does not review your tasks for compliance with target-site terms.
  • Credential ownership. Only connect credentials that are yours or that you are expressly authorized to use. You are responsible for authorized use of every credential you store in the vault, including MFA/OTP codes you relay.
  • No abuse of the platform. Do not attempt to circumvent rate limits, spend controls, custody enforcement, or account freezes, and do not resell the service in a way that hides the end user from us where we are required to know.

We may refuse, suspend, or cancel runs and accounts that violate this section.

4. Billing

  • Every executed attempt bills — including failures and fallbacks. Smart routing can launch up to max_fallbacks additional runners after the primary (default 5) plus budget-governed widening; Learning-family modes fan out to many runners by design. Failed attempts, parallel recovery attempts, and widened attempts are all real executions and all bill.
  • Pricing — runner price plus a 5% platform fee per run (0% while spending welcome credits), plus a $0.10 add-on when the output-normalization pass runs. See Pricing.
  • Spend controlsmax_cost_usd is a hard per-run ceiling in every mode (Learning-family defaults to $20); POST /api/v1/estimate previews cost. At the ceiling no new attempts launch; in-flight attempts finish and bill.
  • Credits — usage is prepaid: you buy credits via Stripe and runs draw them down. Creates are refused with 402 when the balance cannot cover them. Cancelling a run stops new attempts; executed attempts remain billed. Deleting a run does not undo its charges.
  • Refunds[refund policy for unused prepaid credits].

5. Your data

Task inputs, outputs, artifacts, and credentials are handled per the Privacy Policy and the retention contract in the docs. You retain your rights in the content you submit and the outputs of your runs; you grant us the license needed to operate the service (store, process, route to the runners you select or smart routing selects, and produce aggregates/redacted telemetry).

6. Availability and disclaimers

  • No SLA by default. The service is provided without an uptime or latency guarantee; service-level agreements are negotiated Enterprise contract terms. Current platform status is published at /status.
  • No outcome guarantee. Browser automation on third-party sites is inherently unreliable: sites change, block, or challenge automation. A run can fail after billable attempts. Benchmarks and success rates are live measurements, not promises.
  • Third-party dependencies.Runner vendors and model providers are outside our control; their availability, output quality, and data handling are theirs. Session replays and files hosted by a vendor follow that vendor's retention.
  • THE SERVICE IS PROVIDED "AS IS" AND "AS AVAILABLE" WITHOUT WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT, TO THE MAXIMUM EXTENT PERMITTED BY LAW.

7. Limitation of liability

To the maximum extent permitted by law, neither party is liable for indirect, incidental, special, consequential, or punitive damages, or lost profits, revenue, or data; and our aggregate liability arising out of the service is capped at the amounts you paid to Banana Peel in the twelve (12) months before the event giving rise to the claim. [Liability terms pending legal review.]

8. Your responsibility for your automation

You will defend and indemnify Banana Peel against third-party claims arising from your tasks, your use of credentials, or your violation of Section 3 — including claims by target sites relating to automation you directed. [Indemnity terms pending legal review.]

9. Suspension and termination

You can stop using the service at any time. We may suspend or terminate accounts for violation of these terms, non-payment, or risk to the platform, with notice where practicable. Sections 4–8 survive termination.

10. Changes, law, contact

We may update these terms; material changes will be reflected by the date above and, for significant changes, notice in the console or by email. Continued use after a change is acceptance. Governing law and venue: [governing law / jurisdiction]. Contact: [legal contact email].

Command Palette

Search for a command to run...