Clipeus

Architecture

Clean pipeline. No microservices.

Clipeus is intentionally monolithic: one Next.js app, one scan route, one rules directory.

scan pipeline
  1. 01

    Frontend

    Next.js App Router pages — marketing, playground, docs — calling the same API agents use.

  2. 02

    REST API

    POST /api/scan validates with Zod, then hands content to the scanner.

  3. 03

    Scanner Engine

    Normalizes text, runs Unicode inspection, orchestrates rule matching and scoring.

  4. 04

    Rule Engine

    Loads every JSON file from /rules at runtime. Patterns compile to RegExp.

  5. 05

    Scoring Engine

    Dedupes overlapping findings, sums weights, applies action multiplier, emits verdict.

  6. 06

    JSON Response

    Risk score, verdict, categories, flagged segments, explanation, confidence, timing.

Rules as data

Adding a category means adding a JSON file. The loader reads the directory — scanner code stays untouched.

Explainability first

Every match returns the exact substring and a reason string derived from the rule definition.