Applications & Fraud

Application Security

Code, dependency, secret, infrastructure and logic-flaw findings · plain-English requirements · AI agents that fix · developer training

Security agents

Triage agent

Running

Reads every new finding, checks reachability and business context, closes noise with an explanation.

1,204 findings triaged · 71% closed as unreachable

Fix agent

Running

Writes a minimal patch and tests, opens a pull request, and asks the code owner to review.

212 pull requests · 88% merged

Correlation agent

Running

Groups alerts from cloud, code and edge into root-cause issues.

412 alerts → 3 issues today

Threat-model agent

Pilot

Drafts a threat model for each design document and new service.

9 models drafted

Agent pull requests

PR	Repository	Change	Tests	Reviewer	Status
#2291	client-portal	Upgrade XML parser, disable external entities	14 passed	c••• o•••• · portal	Open
#1840	afcos-core	Add requireScope() to STR export	22 passed	c••• o•••• · AFC	Open
#733	infra-terraform	Object lock on evidence bucket	IaC plan clean	p••••• t•••	Merged

Agents work at autonomy level L1: they propose and open pull requests, a human merges. Their tool permissions and limits are on AI Security & Governance.Autonomy levels →