A skilled human inspects the artefact itself — source code, configuration, architecture, policy — and reasons about its security. The finding comes from judgement, not a pattern database.
Systematic, human-driven inspection of an artefact for security weaknesses. The artefact varies; the discipline is the same.
A scanner finds known patterns. Manual review finds:
amount=-100 increases a balanceCommon property: every one of these requires understanding what the system is supposed to do.
Review is most powerful when it happens early. Each stage finds different things; none substitutes for another.
Systematically reason about how a system could be attacked, given its design. Based on Adam Shostack's Four Question Framework:
exec, eval, weak crypto), read surrounding contextInspect the settings an operator chose, not the code itself:
pg_hba.conf, MongoDB binding)Compare against a baseline: CIS Benchmark, vendor hardening guide, or organizational standard.
Read the Terraform / CloudFormation / Helm / Ansible source rather than the deployed result.
Tools: tfsec, Checkov, KICS, OPA / Conftest. They handle easy cases; humans handle architecture and design.
Read diagrams and design documents to find:
Benefits enormously from interactive sessions with system designers.
Read the documents:
The mature stance: automated triages, human decides. Neither replaces the other.
The same shape applies to CSPM + manual cloud review, SCA + manual dependency analysis, and web scanner + manual application test.
False-positive fatigue destroys adoption: 10 high-confidence findings beat 100 noisy ones every time.
Manual review must produce evidence the way every other methodology does.
payment.py accept a client-supplied amount without server-side validation, demonstrated by request X."
Topic 19 covers evidence and traceability in depth.
A client says: "We already run SAST in CI — do we still need manual review?" How do you respond?
SAST is necessary but not sufficient. It catches patterns (injection sinks, hardcoded secrets, deprecated APIs) reliably. It misses authorization bugs, business-logic flaws, race conditions, design weaknesses, and missing controls. Continuous SAST sets the floor; periodic manual review — and threat modeling at design time — provides the ceiling.
Next: Topic 11 — Penetration testing (as methodology). Where manual review ends, controlled exploitation begins.