Application pentesting

Trace an exposed route from request to authorization decision. Turn response differences and source excerpts into a focused test plan.

A valid session and a valid object ID can still produce an access-control flaw. The useful question is where the application binds the requested object to the current identity.

Bring a small, coherent evidence set into the conversation. Ask for the path from input to sensitive operation, the assumptions along that path and the next observation that would distinguish a flaw from expected behavior.

Start with this material

  • A route, handler or request/response pair with credentials and personal data removed.
  • The expected access rule: which role or tenant should own the object and which actions are allowed.
  • Two synthetic test identities and a record of which identity owns each object.

Work through the evidence

  1. Map the input, authentication check, object lookup and response. Ask the model to cite the relevant line or request for each step.
  2. Compare owner, non-owner and unauthenticated behavior. Keep one variable different between requests so a result is interpretable.
  3. Ask for a minimal regression test and a tenant-bound query or policy change. Review the change against the real data model.
  4. Run the selected checks in your test environment. In-product HTTP actions depend on current mode availability, engagement scope and exact-action approval.

A prompt to adapt

Adapt to your evidence
Review the attached report endpoint and these owner/non-owner responses. Find where object access is bound to the session tenant. Separate observed behavior from assumptions. Propose the smallest fix and three regression tests: owner access, cross-tenant denial and missing authentication. Do not treat a successful status code alone as proof of disclosure.

Check the result

  • A claimed disclosure includes the actual non-owner response and identifies the protected data.
  • The proposed fix checks ownership on the server, including alternate routes and indirect lookups.
  • Regression tests fail before the fix and pass after it; error handling does not disclose the object through another field.

Bring your own evidence into a conversation and work through the next question.

Open chat