Is Reffer App safe?

Low risk

Reffer App fetches a server-assigned target URL and CSS-selector schema, then crawls and scrapes up to 500 pages on any site.

The extension periodically polls its own backend for a task that names an arbitrary target URL plus a set of CSS selectors. Using its all-sites host permission, it fetches that URL, extracts matching text, attributes, inline script bodies, links, and images per the server-supplied schema, follows links recursively across up to 500 pages, and uploads the collected content back to the operator's servers. The client enforces no allowlist on which domains can be targeted, so the scope of any crawl is fully controlled by the backend.

palbenalcarazv2.5.7Chrome Web Store
22Risk
Who publishes it

palbenalcaraz - 1 other listing from the same operator, none carrying a finding

What this publisher told the store about itself, and the other listings that told it the same thing.

Publisher
palbenalcaraz

Same store account

1 other listing published from this account, 1k+ users between them, none of them carrying a finding.

AI-generated. Findings may contain errors. Those marked Verified have been manually reviewed.

Publishers can request a review.

Findings

Reffer App fetches any server-named URL and crawls it with no domain limit

MediumunwantedCWE-506Static analysis

Code analysis shows Reffer App's backend can assign a task naming any URL.

What actually happens

You did this

You log into your reffer.ai account in the extension and click Start Training.

This switches the client from idle into work mode and begins polling for tasks.

The extension did this

The service worker fetches a task from reffer.ai's backend and visits whatever URL that task names.

The client places no restriction on which domains the backend can send it to.

What one crawl task can copy off the page you're pointed at

  • Full page text
    Q3 Regional Sales Report, Internal Distribution Only

    Every visible line of text on the page, extracted with the page's own text content.

  • Inline script source
    function initDashboard(){ fetchAccountData(); }

    The complete body of every script tag with no src attribute, including private page logic.

  • Every link and image address
    https://portal.example-target.com/reports/2024-q3.pdf

    The address of every link and image on the page, which also picks the next page to visit.

  • Crawl target address
    https://portal.example-target.com/dashboard

    The exact page the reffer.ai backend told this device to fetch for this task.

The code that does this

The fetch has no domain allowlist; the crawl recurses up to 500 pages

Readable version

Fetching the server-named URL directly, no target-domain check

js/xhttp.js
    processGetUrl: function(url, callback,callbackError){        let method = 'GET';        let contentType = 'html';        let accept = 'text/plain, */*';        let result;        let status = '0';        let d;        let started = (new Date()).getTime()/1000;        let finished = null;        let duration = null;        try {            result = fetch(url, {                method: method,                headers: {                    'Accept': accept,                    'Content-Type': contentType,                    'X-Requested-With': 'Tasker',                },                body: null            });            result.then(response => {                status = response.status;                const r = response.text();                r.then(data=>{                    finished = (new Date()).getTime()/1000;                    duration = finished - started;                    try{                        d = JSON.parse(data);                    }catch (e){                        d = data;                    }                    data = JSON.stringify({'data': d, 'status':status, 'started': started, 'finished': finished, 'duration': duration});                    if(typeof callback == 'function'){                        callback(data);                    }                }).catch(e => {                    console.log('fetch error 1 ['+url+'] '+status, e);                    if(typeof callbackError == 'function') {                        callbackError('getUrlError: '+e);                    }                });            }).catch(e => {                console.log('fetch error 2 ['+url+'] '+status, e);                if(typeof callbackError == 'function') {                    callbackError('getUrlError: '+e);                }            });        } catch (e) {            if(typeof callback == 'function') {                callback('Request error: '+e);            }        }        return result;    },

Recursing into every discovered link, capped only at 500 pages total

workers/domParser/domParser.js
    pushToQueue(item, schema, queue){        if(domParserCounter >= maxPageQty){            log('reached maxPageQty:'+maxPageQty,{queue:queue});            return;        }        log('pushToQueue', {'item':item, 'queue':queue});        const payload = {'url': urlHelper.getUrl(item, this.url)};        if(payload.url === null){            return;        }        schema = this.getSchema(payload.url, schema);        if(schema === null){            return;        }        for(let i in schema){            payload[i] = schema[i];        }        log('pushToQueue: +');        const domParserParams = Object.assign({}, this.parameters);        domParserParams.currentLevel = this.currentLevel+1;        queue.push(domParser(payload, domParserParams));        domParserCounter++;        syslog('Download request '+payload.url);    },

Where task assignment, results and crawl fetches go

    • app.reffer.ai

    Reffer AI's own backend. Assigns crawl tasks via /app/get-tasks and receives the scraped page data via /app/save-result.

    • dashboard.reffer.ai

    Reffer AI's login/authorization page, listed in externally_connectable.

    • operator-chosen crawl target

    The page named in the task payload at runtime. Not fixed in the client and not restricted to any domain list.

Reproduce it yourself

Runs the extension's own page-scraping rules against any saved HTML page, so you can see exactly what a task's extraction schema pulls off a real page.

Requires
  • Node.js 18+
  • npm i jsdom
reffer-domparser-replay.js · js
#!/usr/bin/env node/** * Replays the selector logic from workers/domParser/domParser.js * (queryOne / queryMany / setDataItemValue) against a saved HTML page, * using the extraction schema documented in the extension's own * workers/domParser/readme.MD ("Classic declaration" example). */const fs = require('fs');const { JSDOM } = require('jsdom');// Same rule shapes the extension's own readme documents as a task payload.const schema = {  metaTitle: { queryOne: 'head>title', value: 'innerText' },  metaDescription: { queryOne: 'head>meta[name=description]', value: 'attr', attrName: 'content' },  scriptFiles: { queryMany: 'script:not([src=""])', value: 'attr', attrName: 'src' },  scripts: { queryMany: 'script:not([src])', value: 'innerText' },  links: { queryMany: 'body a', value: 'attr', attrName: 'href' },  images: { queryMany: 'body img', value: 'attr', attrName: 'src' },};// setDataItemValue() from workers/domParser/domParser.jsfunction setDataItemValue(element, rule) {  if (element === null) return null;  switch (rule.value) {    case 'attr':      return element.getAttribute(rule.attrName);    case 'innerText':    default:      return element.textContent;  }}// queryOne() / queryMany() from workers/domParser/domParser.jsfunction queryOne(dom, rule) {  return setDataItemValue(dom.querySelector(rule.queryOne), rule);}function queryMany(dom, rule) {  return Array.from(dom.querySelectorAll(rule.queryMany)).map((el) => setDataItemValue(el, rule));}// processData() from workers/domParser/domParser.jsfunction processData(dom, data) {  const result = {};  for (const key in data) {    const rule = data[key];    result[key] = 'queryOne' in rule ? queryOne(dom, rule) : queryMany(dom, rule);  }  return result;}const file = process.argv[2];if (!file) {  console.error('Usage: node reffer-domparser-replay.js <path-to-saved-page.html>');  process.exit(1);}const html = fs.readFileSync(file, 'utf8');const dom = new JSDOM(html).window.document;console.log(JSON.stringify(processData(dom, schema), null, 2));
How to run it
  1. 1node reffer-domparser-replay.js path/to/saved-page.html

Observation

Static analysis finding. This behaviour was identified by reading the shipped extension code and has not yet been reproduced in a live run. The trigger conditions and the exact data sent are read from the code, not from an observed capture.

What it can do

Permissions this extension asks for, as declared in version 2.5.7. Asking for a permission is not a finding on its own - it is what the extension can do if it chooses to.

  • Read and change your data on every site you visit

    <all_urls> and 2 more

  • Store data in your browser

    storage

  • Act on the current tab, but only after you click the extension

    activeTab

  • Store an unlimited amount of data in your browser

    unlimitedStorage

Where it sends data

Destinations our analysis observed Reffer App contacting. Sending data somewhere is not a finding on its own - an extension that syncs your settings has to talk to its own server - but it is where your data can go, and who else it goes to.

  • app.reffer.ai

    Reffer App sends data to app.reffer.ai. No other extension we have analyzed sends data here.

Updated 2 November 2026cdfdjipcbgankioadglmilfogbknmibn