Skip to content

Acceptable use policy.

The rules for crawling, sending, connecting and using AI with Obelisk.

Effective: November 6, 2026

This policy lists what you may not do with Obelisk. It is part of our terms of service and applies to everyone who uses Obelisk under your account, including the agents you run and the people you invite. "Obelisk", "we" and "us" mean The Obelisk App Inc.

Obelisk crawls websites, connects to your other tools and lets AI agents act for you. That power reaches other people's websites, inboxes and systems, so these rules are strict.

Crawling and audits

Our crawler names itself ObeliskBot/1.0 (+https://theobelisk.app/crawler), reads robots.txt before anything else, honours Crawl-delay, opens public pages only and never logs in. You may not use Obelisk to get around any of that. In particular, don't:

  • Run full or repeated audits of a site you don't own or have permission to audit. A one-off free audit of a public site is fine.
  • Use Obelisk to overload, stress test, attack, scan for vulnerabilities in, or probe any site, server or network.
  • Point Obelisk at private networks, internal addresses, sign-in pages or anything behind a login, or try to make the crawler reach them.
  • Get around a site's robots.txt, firewall, rate limits or bot protection, or ask us to whitelist a site you don't control.
  • Add a site to more than one workspace or account to multiply crawl limits.

Email, messages and tickets

Agents can send email and post to Slack, Linear and Jira for you. Obelisk sends email only to people who confirmed they want it (double opt-in) or to a team member's own verified address, and limits how much any workspace sends. You may not get around those controls. In particular, don't:

  • Send spam or unsolicited bulk or commercial messages, or add people to the email list who haven't agreed to hear from you.
  • Import bought, rented, scraped or harvested addresses, or confirm a recipient on someone's behalf.
  • Send messages that are deceptive, impersonate someone, hide who sent them or how to stop them, or break anti-spam laws such as CAN-SPAM, the UK and EU e-privacy rules or CASL.
  • Use email, Slack, Linear or Jira actions for phishing, harassment or malware.
  • Flood a channel, project or inbox with automated posts.

Connected accounts and agent actions

You can connect WordPress, GitHub, Slack, Linear, Jira, Google Search Console and Google Analytics. Agents act on them only when you approve an action or when the autopilot settings you chose allow it. Don't:

  • Connect an account, site, repository or workspace you aren't authorised to connect, or use someone else's credentials.
  • Use agents to publish or change content you don't have the right to publish or change.
  • Set up autopilot to take actions that break this policy, the law, or the rules of the connected service.
  • Use Obelisk to break the terms of a connected service, such as Google's, GitHub's, Slack's, Atlassian's or Linear's.

Prompt injection and AI misuse

Obelisk's agents read web pages, your uploads, chat messages and data from connected tools. Some of that text may try to give the agents orders. Don't:

  • Write or plant instructions in a website, document, upload, chat message, ticket or any other input that Obelisk reads to make an agent ignore its rules, reveal its instructions, reveal secrets or other people's data, change its approvals or take an action nobody approved.
  • Plant such instructions in a site so that other Obelisk customers' audits or agents act on them.
  • Try to jailbreak Obelisk's AI, get around its safety rules, spending caps or approval steps, or make it produce content this policy forbids.
  • Use Obelisk to produce content that is illegal, sexual content involving minors, violent extremist material, targeted harassment, or deliberately false information meant to deceive people.
  • Mass-produce pages, reviews or links meant to deceive search engines or AI answers, such as fake reviews, doorway pages or link schemes.
  • Use Obelisk's outputs or behaviour to train, benchmark or build a competing model or product.

If you find a way to make an agent misbehave, tell us at support@theobelisk.app instead of using it. We appreciate those reports.

Tracking on your websites

If you add the Obelisk tracking script to a website, don't:

  • Put it on a website you don't run or have permission to measure.
  • Run it without a cookie banner where the law requires consent, or without describing it in your privacy policy. You can use this wording.
  • Map form fields other than email address and name for lead capture, or send passwords, payment card numbers, government ID numbers, health, financial or other sensitive data, or children's data, through identify(), custom events or form fields.
  • Put it on a website or page aimed at children under 16.

Uploads and content

Don't upload, store, publish or share through Obelisk anything that:

  • Is illegal, infringes someone's copyright, trademark, privacy or other rights, or that you don't have the right to share.
  • Contains malware or code meant to harm, or that tries to attack Obelisk or anyone else.
  • Contains special category data (health, biometrics, religion, sexual life and similar), payment card data or government ID numbers, unless Obelisk needs it for a task you asked for and the law allows it.

The service itself

Don't:

  • Create several accounts, use disposable email addresses or automate sign-ups to get more free audits, trials or credits.
  • Get around plan limits, quotas, bot checks, rate limits, spending caps or access controls.
  • Access or try to access another customer's workspace or data.
  • Test Obelisk's security, scan it, or load test it without our written permission. Report security issues to support@theobelisk.app.
  • Copy, decompile, reverse engineer or rebuild Obelisk, or scrape it, except where the law allows despite this rule.
  • Resell, sublicense or bulk export Obelisk or the third-party data in it, except as your plan allows. Agencies may share reports with their own clients.
  • Use Obelisk to break any law, including sanctions and export laws.

What we do about it

We may remove content, stop an action, pause agents, block a site, limit sending, suspend or close an account, or report activity to the authorities, depending on how serious it is. We try to warn you first and give you a chance to fix it, but we act right away when there is a risk to people, to other sites or to Obelisk. Breaking this policy is a breach of our terms.

To report abuse of Obelisk, including ObeliskBot crawling your site or a message sent through Obelisk, email support@theobelisk.app with the details.

Changes

We may update this policy as Obelisk and its risks change. We'll tell account owners about changes that matter at least 14 days before they take effect, unless a change is needed sooner to stop abuse or comply with the law.