For Auditors
This page is for external auditors reviewing a company that runs its compliance in Kantis. It covers what your access includes and how to point your own AI tool at the customer’s records.
Your access
Section titled “Your access”The customer invites you to their Kantis organisation with the Auditor role. That role is read-only: you can see the customer’s records but not change them. You sign in at app.getkantis.com with the email they invited.
You can read the customer’s policies (with version history), evidence tasks and their attachments, evidence documents, frameworks and controls, the Statement of Applicability, the risk register, vendors and people.
If you work with more than one Kantis customer, each is a separate organisation. Switch between them from the organisation menu.
Connect your AI tool
Section titled “Connect your AI tool”You can connect Claude, ChatGPT or any MCP-capable assistant to the customer’s organisation and ask it questions about their records. Follow the steps on Connect AI Agents (server https://mcp.getkantis.com/). Before you approve, make sure the customer’s organisation is the active one in Kantis: the connection is bound to it. The approval page shows the connection as read-only auditor access.
An auditor’s connection is read-only and is set up for audit work:
- Per-requirement view for ISO 27001.
list_requirementslists every clause (4.1 to 10.2) and Annex A control (A.5.1 to A.8.34) with the SoA applicability, how many linked policies are published, how many evidence tasks are done, and evidence form submissions.get_requirementshows one requirement in full: its SoA row and justification, the policies that describe it, the evidence tasks with completion dates and attached file names, and the evidence forms. - People evidence.
list_people_complianceshows, for each person in the customer’s workforce, which required policies they have accepted, the security awareness training they completed and when, and their device security (Kantis Device Agent checks or reviewed screenshots). Leavers are listed separately. - The usual records. Policies (with how many people accepted each), tasks, risks (with the treatment reviewer and linked tasks), vendors (with their risk assessment), people (including leavers), documents, findings, the SoA and the audit trail, each with record IDs and timestamps to cite.
- File contents, if the customer allows it. With the customer’s permission,
read_evidence_filegives your assistant the text of attached documents (PDF, Word, Excel as CSV) and screenshots as images. Never a download link. - Not included. The customer’s sales security questionnaires and Trust Center access requests.
Everything your assistant reads is recorded in the customer’s audit trail under your name. Connections expire after 30 days; the customer can disconnect them at any time from Settings → AI Agents.
Questions to try
Section titled “Questions to try”- “List the Annex A controls marked not applicable in the SoA, with the justification for each.”
- “Show me the evidence for A.5.15 Access control and read the attached files.”
- “For clause 9.2, what internal audit evidence is there and when was it completed?”
- “Which requirements have no published policy or no completed evidence task?”
- “Who has not accepted the Acceptable Use policy or finished security awareness training?”
For the customer: allowing file reading
Section titled “For the customer: allowing file reading”File reading is off by default. An organisation owner or admin turns it on in Settings → AI Agents with Let AI agents read evidence files. It applies to every AI connection in the organisation, including your auditor’s. Each file read appears in the audit log with the person, the file and the AI tool, and you can turn it off again at any time.