Attack library7 published write-upsAll six categories

Attack library.

MATT runs 1,200+ attack patterns across six categories. These are the ones written up in full: what the pattern is, where it hides, how it is proven against a live product, and the rule and executable check that keep it closed.

  1. AP-0104Identity & access Object ownership missing on a generated endpoint An endpoint accepts an object identifier from the caller and returns or changes the object without confirming the caller owns it. It is the most common finding in code written by assistants, which copy the shape of neighbouring handlers but not their authorisation. Severity · High Open pattern →
  2. AP-0112Identity & access A revoked session that keeps working A user logs out, changes their password or is removed, and a token issued earlier keeps working. The product treats a valid signature as proof of a valid session, so a credential that should be dead is still accepted. Severity · High Open pattern →
  3. AP-0206Spend & resource abuse Metered features a stranger can spend for you A product spends money on each request it serves: model calls, compute, storage, messages. When a metered feature is reachable outside the flow that meters it, a stranger can run up the bill and the meter never moves. Severity · High Open pattern →
  4. AP-0308Data exposure An API response carrying more than the screen shows The interface shows a name and an avatar; the response behind it also carries the email, the internal identifiers, the role flags and a token. The screen hides the extra fields, and the network tab reveals every one. Severity · High Open pattern →
  5. AP-0403Automation & fraud A reset flow that reveals which accounts exist A password reset form answers differently depending on whether the address has an account. The difference lets an attacker sort a list of millions into real customers, then aim the rest of the attack only at them. Severity · Medium Open pattern →
  6. AP-0502Availability One cheap request that stops the product A single request that is cheap to send and expensive to serve can slow or stop the product for everyone. An unbounded search, an oversized upload or a deep nested payload does more damage than any flood. Severity · High Open pattern →
  7. AP-0605AI-written & agentic code Content an agent reads becoming a command it runs An agent reads a document, an issue or a web page as part of its task, and the content contains an instruction. The agent cannot tell the data from the command, and it acts on both with the permissions it was given. Severity · Critical Open pattern →

The rest of the catalogue

Every pattern MATT runs belongs to one of six categories. The write-ups here are published as the research team finishes them; the catalogue itself runs against your product from the first week, whether or not its entry has been written up.

  • Identity & access2 published
  • Spend & resource abuse1 published
  • Data exposure1 published
  • Automation & fraud1 published
  • Availability1 published
  • AI-written & agentic code1 published
Read the rule set → See the six categories and what each one covers →
ExploitOps discovery call

Understand where your security loop breaks.

A 15-minute call with a MATT co-founder to understand your product and whether an ExploitOps Review is worth doing. If there is a real need, we scope one with the research team.

Book a discovery call→