Dregs Manual
Dregs is a fraud detection and abuse prevention platform for web applications, particularly SaaS businesses. It continuously scores your users across four dimensions (Humanity, Authenticity, Uniqueness, and Behavior) so you can identify bots, freeloaders, spammers, and bad actors before they do damage.
What You'll Find Here
This manual covers everything from initial integration to fully autonomous fraud response. Whether you're adding the tracking script for the first time or configuring webhook-driven automation, you should find an answer here.
Please email dregs@dregs.com if you don't find what you're after!
New to Dregs? Start with How Dregs Works to understand the approach, then follow Getting Started to integrate in minutes.
Getting Started
- How Dregs Works: The problem Dregs solves, the four scoring dimensions, and the path from manual review to fully autonomous abuse prevention.
- Getting Started: Add the tracking script, identify users, send events, and verify data and scores in your Dregs dashboard.
Core Concepts
- Tracking: Frontend tracking via dregs.js and backend tracking via a server SDK or the REST API.
- Identities: How users become scored profiles with devices, events, relationships, and badges.
- Groups: The organizations, teams, and workspaces your users belong to, and how Dregs records membership.
- Devices: How Dregs tracks devices, maps device-to-identity relationships, and uses device intelligence for scoring.
- Events: Tracking user actions and feeding behavioral data into the scoring engine.
- Scoring: The four scoring dimensions explained: what they measure, how they work, and what the numbers mean.
Features
- Dashboard: Navigating the Dregs dashboard: identities, devices, events, scores, and search.
- Escalations: Configuring escalation rules that open stateful incidents when scores cross thresholds.
- Badges: Automatic labels that classify users, from your score rules and from Dregs's analyzers.
- Gates: Ask Dregs whether an identity should take an action right now, and get allow, review, or deny.
- Channels: Delivering fraud notifications via email, Slack, and the dashboard.
- Webhooks: Automated fraud response via HMAC-signed HTTP callbacks to your application.
Advanced
- Datasets: Built-in lists of disposable, free, and suspicious domains and IPs, plus your own entries and exclusions.
- Mappings: Telling Dregs what your event names and identity fields mean, so analyzers recognize them.
- Team Management: Inviting team members, managing roles, and configuring API credentials.
- Limits and Quotas: Plan allowances, API rate limits, payload caps, webhook retries, and what happens at each cap.
- Server SDKs: Official clients for Python, Node, Java, Ruby, PHP, Rust, and Elixir: send backend events and read scores without writing your own HTTP layer.
- REST API: Querying scores, identities, and devices programmatically from your backend.
- AI Agents: Connecting Claude, ChatGPT, Claude Code, Codex, Cursor, VS Code, and other assistants and agents through the Dregs MCP server.