File DA-009 View as Markdown

Core

The detection engine inside the script is open source: github.com/doubleagent-so/core (MIT). Read how every verdict is made, run it yourself, or make it better.

• Runs in the page. No dependencies, no network requests.
• Returns human, bot or agent, with probabilities, the agent's identity, the reasons and a recommendation.
• 226 named bots and agents in the catalog, 71 signal codes.

Script or core

The script runs core. With data-cloud="off" it stays in the page, like core on its own.

Use it

git clone https://github.com/doubleagent-so/core && cd core && npm ci
npm run example   # live verdict page

import { createEngine } from '@doubleagent-so/core';

const engine = createEngine(window, { onVerdict: (v) => console.log(v.class, v.reasons) });
const verdict = await engine.ready;

Getting started covers the options, the verdict and sending engine.payload() to your backend.

How it works

Probes and input events become signals. Each is a log-likelihood ratio in a capped group, so correlated tells can't pile up. Fusion adds them to your site profile's priors. A policy turns the result and the action into a recommendation. How it works.

The cloud re-scores every session with evidence the browser can't forge (headers, IP ranges, Web Bot Auth). It never trusts the client's verdict.

Improve it

• A real browser labelled automated, or an agent that passed: report a misclassification.
• A new agent or bot: add it to the catalog.
• A probe, better weights, fewer false positives: Extending and Contributing.

Found a way past it? Report it privately (Security).