Privacy
What happens to what you send us.
Short and specific, because there is not much to say. Last updated 16 November 2026.
Who we are
Chain Defenders is a two-person security and AI consultancy. For anything in this notice, write to audits@chaindefenders.xyz.
What we collect
Only what you send us, through one of five forms:
- The audit request: project name, repository link if you give one, chains, rough size, timeline, your name if you give it, an email address or Telegram handle, and any notes you write.
- The price estimate: what you tell us about the code (its size, language, what it does, whether it has tests, and an after-audit plan if you pick one), the project name, the repository link if you give it, your name if you give it, an email address or Telegram handle, and the estimate we showed you. If you give an email address, we email you the estimate.
- A grant application: your project or company name and website, what you tell us about it, your timing and team, your name if you give it, and an email address or Telegram handle. For an audit grant also the repository and docs links, chains, language, size, stage and TVL if it is live; for an AI grant your field, team size, monthly AI spend or the hours the job takes, the providers or systems you use, and who would own the tool. It is also saved as a request in the matching practice, with the programme as its campaign tag, and we keep the score and decision we record on it.
- A referral: your name and email address, the project's name, website or X account and what it needs, how you know them, and the name and email address or Telegram handle of the person there you introduce us to. If the project signs, we also keep where you want your commission sent and the commission payments we make to you. We use the project's contact details only to introduce ourselves about your referral; tell them before you send them to us.
- The AI review request: company name, website, team size, industry, what you spend on AI or the hours you spend on a task, the tools you use, your name and role if you give them, your email address, budget, timeline and what you tell us about the problem.
If you write to us another way (Telegram, email, an introduction or a conversation at an event), we may save it as a request by hand: the project or company name, the website and repository if you shared them, your name and your email address or Telegram handle, where you reached us and a short note of what you asked for. It is kept and protected like a request from a form.
If we are simply in touch with you about something else (a partnership, a question, an introduction), we may keep the conversation in our admin instead: your name or your project's, your email address or Telegram handle (encrypted), a short note of what it is about, and the messages we exchange (encrypted). We keep it while we are in touch and delete it when it is no longer needed, or sooner if you ask.
With any of them we also store a one-way hash of your IP address and your browser's user-agent text, to spot abuse. If you arrived through a link with campaign tags (utm_ parameters), we keep those on the AI request.
On the AI page, the calculators and the self-check record anonymous usage events, such as "calculator used", with that same hash. They contain no text you type and nothing that identifies you.
What we do not do
- No analytics service, no advertising or tracking scripts, and no cookies.
- The self-checks and calculators run in your browser. Nothing you answer is sent to us until you press the button to send a request.
- Your browser's session storage remembers only that you dismissed a banner. It is cleared when you close the tab.
- We do not sell your data, and we do not share it for marketing.
Where it goes
- Our database. Your name and your email address or Telegram handle are encrypted before they are saved. The rest of the request is stored so we can answer it.
- A private Telegram chat. When a request arrives, a message with its details, including the contact detail you gave us, is sent to a private chat that only we can read, through Telegram's bot service. We also mark the request's progress (for example “contacted”) on that message from the chat, and once a day the chat gets a reminder of requests we have not answered yet, by project or company name only. Telegram therefore processes that message under its own terms.
- An AI model, to help us write. When we reply to you or follow up, we may give an AI model (Groq, or Cloudflare when Groq is busy) what you asked for and what we have written to each other, with email addresses, phone numbers and handles taken out, to draft our message. We read and edit every draft, and a person sends it.
- Our hosting. The site and its database run on our hosting providers' infrastructure, which handle the data on our behalf.
Why, and how long
We use your request to reply to you and, if you go ahead, to scope the work. We rely on your consent, which you give by ticking the box on the form. We keep a request for as long as we need it to reply and to keep a record of the enquiry, and we delete it sooner if you ask. With it we keep a copy of what we write to you and, when we add them, your replies (encrypted), so the conversation stays in one place; they are deleted with the request.
Businesses we contact first
We also look for companies and small businesses we might be able to help, using public sources: listings of crypto protocols and their exploits, public code on GitHub, public governance posts and grant applications on ecosystem forums, job posts on Hacker News and on job boards, product launch listings (Hacker News and the Y Combinator company directory), startup funding news, and open map datasets (OpenStreetMap and Overture Maps). From what they publish we keep the business's name, website, public social profiles, location, what it says about itself, and the contact address it publishes, which can be a person's own address when that is the one they put out. We encrypt email addresses, and we never take anything from behind a login.
We use this to decide whether to write to them, and then to write. Our basis is our legitimate interest in offering our services to businesses that appear to need them. In our first message we say where we found them and how to opt out. To draft that message we may give an AI model (Groq, or Cloudflare when Groq is busy) the public facts about the business, never a contact address, and for a later message what we have written to each other, with contact details taken out; we read and edit every draft, and a person sends it. We keep a copy of what we write and of any reply (encrypted), deleted with the rest. We do not use this for anything else, and nothing is sent automatically.
If you ask us not to contact you, we delete what we hold and keep only your website's domain, your public handles, the public listings we found you in, and a one-way code of your email address, so our search never finds you again. Anything we have not acted on is deleted after twelve months.
If you accept an offer
An offer we send you comes with a private link. If you accept it there, we keep your name, your role, your email address (encrypted) and the time you accepted, as the record of the agreement, and our private Telegram is told that you accepted, by name. We keep this for as long as we keep the records of the work, as the law on accounting asks.
If you sign an NDA with us
A non-disclosure agreement we send you comes by email with a private link. If you sign it there, we keep your name, your title, your email address (encrypted), any company details you complete when signing, the time you signed and a fingerprint of your internet address (not the address itself), with the agreement as it was sent, as the record of the signature. We keep the email address we sent it to (encrypted) and the company details on the agreement. The agreement is stored in our private GitHub repository, the signed copy is emailed to you and to the address we sent it to, and our private Telegram is told that you signed, by name. We keep this for as long as the agreement and its confidentiality period last, and then as long as we may need it to show what was agreed. If we sign an NDA on your company's paper instead, we keep a copy of it, with the names of those who signed it, for the same time.
If we invite you to the client portal, or you made a referral
When we invite you to follow a project in our client portal, we keep your email address (encrypted), your name if we have it, who invited you and when you last signed in. To sign in, we email you a link that works once (within 7 days for the one in an invitation or a referral receipt, 15 minutes for one you ask for); we keep only a fingerprint of it. Signing in sets one cookie that keeps you signed in on that device and does nothing else; it ends after 90 days without a visit, or when your access is removed. A referrer signs in the same way, with the address they referred with, to see where their referrals stand and what they have earned. When you reply to a finding, your reply and your name are posted on that finding in the private GitHub repository where we keep your project's findings, and our team is told in our internal chat; your email address is not. When we share something new with you, we may email you a short note saying so, at most one an hour; reply and we stop. We remove your access when the project no longer needs it, or when you ask.
If you write us a testimonial
When your project's final report or handover is shared, we may ask you in the portal (and by email, once, with one reminder) for a few words about working with us. If you write them, we keep your words, the name, role and company you give, any links, and the picture you choose, together with who sent it. Nothing is shown until we have read it; we may tidy typos, and we keep what you wrote as you wrote it. Approved words appear on this site and may be shared in our posts on X and LinkedIn. Ask us and we take them down.
If your project has a badge
When a project we worked on qualifies for a badge, we issue one: a page at its own random address on this site shows the project's name and website, what the badge certifies (for example the Security Grade of the final report, the commit we reviewed, or a saving as a share of the AI bill, never an amount) and whether it is still valid, with those facts signed by us. The page is not listed or indexed anywhere; it is found through the badge, so it is public once you show the badge. Badge images are served from this site, which sees only what any image request carries. Ask us and we revoke a badge; its page then says so.
Your rights
You can ask to see what we hold about you, to correct it, to have it deleted, or to withdraw your consent at any time, by writing to the address above. If you are in the EU or UK you can also complain to your data protection authority. Because we store only a hash of your IP address, we may need you to tell us the email address you used so we can find your request.
Changes
If what we collect or who handles it changes, we will update this page and its date.