One engine. Two layers.
Every output checked.
Reglint enforces compliance at both development time and runtime — so violations never reach production, and never reach your users.
Layer 1 · Code Analysis
Before you ship.
A GitHub Action scans your code on every pull request for hardcoded secrets, exposed PII, and unsafe AI prompts across 50+ regulation patterns — so violations never reach production.
Each flagged line carries the rule name and the citation. Your engineers see it in the PR diff, not in a compliance audit six months later.
Layer 2 · Agent Monitor
At runtime.
Reglint sits between your agent and the user. Every output is checked in real time and returned as BLOCK / REDACT / ALERT / PASS — each backed by a specific citation.
One POST to /api/monitor/scan. Sub-second. No change to your agent logic.
Integrations
Drop into the stack you already use.
One POST to /api/monitor/scan — or native hooks for the tools below.
Where this is going — Reglint is the compliance layer today. The plan is to become the legal infrastructure for AI: a layer that sits across your entire agent stack, enforcing, citing, and auditing every decision. Your data stays yours — we don't train on customer data by default.