For your IT and security team

Written to survive a security review, not to dodge one.

You are being asked to let a tool near your registration data and to accept that a language model reads your study reports. Both deserve a specific answer rather than a reassuring one. Here is what is actually in place, and what we do not have.

What we do not hold

The shortest part of any review is the attack surface that does not exist.

  • No IUCLID URL, hostname or IP address for your installation
  • No IUCLID account, password, API key or session token
  • No inbound connection to your network, no firewall rule to open, no VPN
  • No card data. Payment runs on Stripe's hosted pages, never through our forms

The application

Standard controls, applied everywhere rather than where it was convenient.

  • Passwords hashed with bcrypt. Session, invite and pairing tokens stored only as hashes
  • Session cookies are HttpOnly and SameSite. CSRF protection on every authenticated form
  • Content Security Policy, HSTS, frame ancestors denied, MIME sniffing off
  • Every record carries its organisation and every query is scoped to it. Email verified accounts, owner and member roles
The question everyone asks

The model reads. It does not act, and it does not decide.

It has no route to IUCLID

The language model receives the study report and the field schema, and returns structured data. It has no network access to your systems, no credentials, and no way to trigger an action. What it produces is text until a person approves it.

Two gates stand after it

Its output is validated against the IUCLID field schema for that endpoint, and anything outside the schema is reported rather than written. Then a qualified person reviews and approves. Only an approval produces a plan the extension will execute.

Your content is not training data

Drafting runs on Anthropic's commercial API, under terms that do not use your inputs or outputs to train models. Source reports are deleted from our storage once every endpoint they were tagged to has been resolved.

The extension

Seven permissions, each with a reason your reviewer can check.

PermissionWhat it is for
activeTab Access to the one tab you clicked on, only on that click. This is what replaces a standing permission over your IUCLID address, which we deliberately do not hold.
scripting Injects the two scripts that find the record and fill the form. IUCLID offers no write path the extension could use instead, which is the whole reason the form is driven directly.
storage Holds one value: the token that authorises this browser to fetch your organisation's approved jobs. It never leaves the browser except to your own service.
sidePanel The whole interface is a panel docked beside IUCLID, opened only by your click on the toolbar icon. It replaced a pop-up, which closed the moment focus moved and took a run in progress with it. Writing a record takes minutes, and you need to see which field is going in and be able to stop it.
notifications One notification, raised only when a run finishes while the panel is closed, saying whether the record was written or needs you. If the panel is open, nothing is raised. Never used for marketing, prompts or re-engagement.
alarms One recurring alarm at Chrome's one minute minimum, as a backstop for the background worker Chrome is entitled to suspend. It gives the extension a legitimate reason to wake and resume watching a run it has already started. It starts nothing on its own.
debugger Used for exactly one action. IUCLID's "add new document" menu ignores scripted clicks, so one genuine click is dispatched at that one control and the debugger detaches immediately, with Chrome's own banner visible while it does. It is never used to read page content, network traffic or console output, and never attached to another tab.
  • No host permission is granted at install. It asks for one origin during pairing, your Dossier Flow service, and one more the first time you run a job: your own IUCLID address, through Chrome's own prompt, once. Nothing else is ever requested.
  • No remote code. Every script ships inside the reviewed package, readable and unobfuscated, and nothing is fetched or evaluated at runtime.
  • No analytics, no advertising, no tracking, and no access to any site other than the IUCLID tab you clicked on.
  • Deployable centrally. Your administrators can push it by policy rather than asking each user to install it.

Send us the questionnaire.

If your organisation runs a formal vendor assessment, we would rather answer it in writing, in this much detail, than have you take any of the above on trust. Send the form and we will return it completed, with the code references behind each answer.