Step 1
Map which workloads will send supported traffic through the gateway.
Security
Security starts with a clear gateway boundary
Security teams need to know what the gateway enforces, what traffic is in scope, and how to report issues. This page summarizes the product boundary and points to trust and documentation resources for deeper review.
What you are trying to solve
Shadow AI integrations can create unaudited paths to external models.
Step 2
Deploy inside the network boundary your security team approves.
Step 3
Validate policy tests and evidence sampling before mandatory routing.
Security topics covered on this site
Verdictan is a self-hosted AI governance gateway for supported Chat Completions, Responses and Messages traffic. Unsupported request families remain outside the documented enforcement boundary until explicitly supported.
Decision events help security teams reconstruct allow, redact, block, and escalation outcomes for supported requests. They do not replace your SIEM, ticketing, or incident response process.
Resources for security review
Trust information
Find legal, privacy, and security contacts on the trust page.
Security.txt
Use the RFC 9116 security contact published at /.well-known/security.txt for coordinated disclosure.
Policy testing
Run deterministic policy tests in staging before you require production traffic to use the gateway.
Questions to answer during evaluation
- Which environments and provider targets are approved for your data classes?
- Who owns gateway configuration changes and credential rotation?
- How will you sample decision evidence during incidents and audits?
Related paths
Continue with a template, product area, or guide that matches this page.
Supported traffic uses a reviewed gateway path with testable policy and retained evidence.
Read trust information and report security issues through the documented contact path.