Field notes on agentic offensive security
Deep, practitioner-written pieces on agentic penetration testing, proof-based validation, and how autonomous testing changes offensive security. No fluff, no gated PDFs.
BOLA in GraphQL: how ‘opaque’ Global IDs still leak data
GraphQL hands the top API flaw, broken object level authorization, a bigger and harder-to-scan surface. How it breaks behind base64 Global IDs, three disclosed proofs, and the two-account query that proves it.
Read the article → API SecurityBOLA vs IDOR: same bug, different name?
IDOR and BOLA describe the same broken-authorization failure from two eras of security. The difference in the name, why UUIDs don't fix it, why scanners miss it, and the two-account request that proves it.
Read the article → API SecurityAPI authorization is the #1 risk of the agentic era
AI agents are non-human identities calling APIs at scale, on a protocol with no authorization model. Why BOLA and BFLA are the one flaw scanners, WAFs, and autonomous pentest tools cannot test, and how to close the gap.
Read the article → API SecurityAPI security best practices for 2026: the checklist that maps to real breaches
Most API security advice guards the perimeter; the breaches come through broken authorization. Twelve OWASP-mapped controls, a copy-paste checklist, and how to test the ones scanners cannot see.
Read the article → Security ResearchOne resident login, an entire apartment complex: the master PIN in Rently's API
The app showed us two codes; the API returned the property's full code table, master PIN included. The story behind CISA advisory ICSA-26-237-01, and the lesson it holds.
Read the article → API SecurityBOLA vs BFLA: the API authorization flaws scanners miss
Authorization, not injection, breaks most modern APIs. The difference between object-level and function-level authorization, why scanners find neither, and how to test for both.
Read the article → ComparisonAgentic pentesting vs DAST: what's the difference?
DAST matches patterns on one app and hands you unverified alerts. An agentic pentester reasons across your whole surface, chains findings, and proves exploitability. How they differ, and when to use each.
Read the article → MethodHow agentic pentesting proves an exploit
Inside proof-based validation: a step-by-step look at how a leaked token becomes a proven, CVSS-rated critical finding, and why reproducing it before reporting changes everything.
Read the article → MethodHow to pentest an LLM or AI agent
Testing the model, its tools, and its data as one system: prompt injection, tool abuse, and data exfiltration, mapped to the OWASP LLM Top 10.
Read the article →Point the agent at your attack surface
Give us a domain and the rules of engagement. We will return a scoped run and show you what it surfaces, and what it proves.