Responsible disclosure

Last updated: September 6, 2026

Esan acts on your behalf: it signs into your apps, sends email, and can spend money within limits you set. A vulnerability here is not an abstraction, and we would rather hear about one from you than from a customer. If you have found something, this page tells you how to tell us and what happens next.

How to report

Email security@esan.ai. Include enough for us to reproduce it: the URL or endpoint, the steps, what you expected and what happened. A short screen recording is worth more than a long description. Write in English or Spanish, whichever you prefer.

Please report to us first and give us a chance to fix it before discussing it publicly. We will not ask you to stay quiet indefinitely — see the timelines below.

What we promise

• An acknowledgement within 3 working days. From a person, not an autoresponder.
• An assessment within 10 working days — whether we consider it a vulnerability, and if so, roughly when we expect to fix it.
• We tell you when it is fixed, and you are welcome to verify it.
• Credit if you want it. Tell us the name or handle you would like used. You are equally welcome to stay anonymous.
• No legal action for research conducted under this policy. See the safe harbour below.

There is no bug bounty. We are a small team and we would rather say so plainly than imply a reward we cannot pay. If that changes, it will be announced here first.

Ground rules

Research under this policy has to stay within these limits. They exist to protect other people's data, not to make your work harder.

• Use your own account and your own data. Do not access, modify or exfiltrate anyone else's chats, files, wallet or connected apps. If a bug gives you access to someone else's data, stop, keep no copy, and tell us what you saw.
• Do not degrade the service. No load testing, no denial of service, no mass automated scanning against production.
• No social engineering of our staff, users or suppliers, and no physical intrusion.
• Do not use a finding beyond proving it. Reading one record to demonstrate access is proof; dumping the table is not.

Safe harbour

If you follow this policy in good faith, we will treat your research as authorised under the Spanish Criminal Code and the EU Directive on attacks against information systems, we will not pursue civil or criminal action against you, and we will not report you to law enforcement. If a third party brings action against you for research you conducted under this policy, we will make it known that your activity was authorised.

If you are unsure whether something is in scope, ask before you test. An email to security@esan.ai costs you a day and settles it.

In scope

• The web application at esan.ai and its API routes.
• The agent backend that runs tasks.
• Anything that lets one account reach another account's data, sessions, connected apps or wallet.
• Anything that lets input from a web page, a file or another person's agent make Esan act outside what its owner authorised — prompt injection with a real consequence, not a screenshot of a model saying something odd.
• Authentication, session handling and the sign-in code flow.

Out of scope

These are either not vulnerabilities or not ours to fix, and reports about them will be closed without much ceremony:

• Automated scanner output with no demonstrated impact, and missing hardening headers on their own.
• Findings in third-party services we use (Stripe, Supabase, Vercel, model providers). Report those to them; tell us too if it affects Esan users.
• Model behaviour that is unhelpful, biased or wrong but has no security consequence. That is feedback, and it is welcome at support@esan.ai.
• Anything that requires a compromised device, a rooted browser, or the victim pasting attacker-supplied content into their own settings.
• Denial of service, volumetric or otherwise.
• Reports generated by scanning our public marketing pages.

Coordinated publication

We would like a fix to be live before a finding is public. Our default is 90 days from your report, or sooner if we fix it sooner — we will tell you when we ship. If we go quiet or the 90 days pass with no fix and no explanation, publish. A researcher should not have to keep a company's secret indefinitely because that company was slow.

If a vulnerability is being exploited in the wild, tell us immediately and we will work to a shorter timeline with you.