Resources

Three things you can run yourself.

We built all three for our own work and left them open. One checks a consumer MCP server against a security baseline we wrote. One tells you what actually broke in AI security this week. The third tells you how far your DPDP programme still has to go.

None of these is a red team. They are questionnaires and feeds — you run them, they take minutes, and they work from what you tell them. Our actual work is adversarial: we attack the running system and report only what we could make happen. A self-assessment tells you which questions to ask. A red team tells you which answers are true.

None is a scoreboard with a sales call attached. The briefing needs no account at all. The MCP baseline keeps your answers in your browser unless you ask for the report; it ends by saying whether a red team is worth it for you, and the verdict stands without one. The readiness check stores your name, your organisation and your work email — never your answers — and hands back a list of what to fix, whoever you then hire to fix it. We say elsewhere on this site that a tester who also sells the remedy has a reason to find something; that applies to what we give away too, so these are built not to.

Field guides