AI Output Used as Intent / URL (navigation & redirection)

← AI + Mobile (LLM x IPC / WebView)

ai_output_to_intent_url   MEDIUM

Description

The assistant’s output is fed directly into startActivity()/url launcher, so a prompt-injected model can drive navigation, open redirects, or fire intents on the user’s behalf without a confirmation boundary.

How it works

The assistant output, attacker-steered via the prompt, is fed directly into a URL launcher / startActivity() with no allowlist and no user confirmation. A prompt-injected model can drive navigation to an attacker host, an open redirect, a javascript: URI, or a privileged deep link on the user’s behalf. The demo drives a model and navigates a real WebView to the model-chosen URL. The secure path requires both an https host allowlist and explicit user confirmation, so model output alone cannot navigate the app.

How to exercise it. DVMA is the harness - open this module from the home index and tap the demo action. The screen ships the malicious input and simulates the attacker (e.g. the companion app, crafted intent, or scanned payload) in-process, and the evidence panel prints the proof. The Tools (optional) and Attack inputs below are only needed to reproduce the exploit end-to-end on a real device.

Exploit steps

1. Set up. Build DVMA with a flavor that enables the AI + Mobile (LLM x IPC / WebView) category (e.g. --dart-define-from-file=config/flavors/dev.json) and run on an emulator/simulator you control. The demo needs no external tooling; for the optional on-device reproduction the relevant tools are: garak, promptfoo, frida · Android: adb · iOS: xcrun simctl openurl.
2. Locate the target. From the home index, open AI Output Used as Intent / URL (navigation & redirection) (ai_output_to_intent_url). The How it works section above describes this module’s specific weakness; the screen states the intended-secure behavior and exposes the vulnerable action.
3. Exploit. Deliver attacker content across the mobile boundary (deep link / clipboard / QR) into the assistant, or steer the model’s OUTPUT into a WebView / intent / tool call, and confirm it executes with no validation boundary in between.
4. Observe the evidence. Trigger the vulnerable action and read the evidence panel - it prints the concrete proof (leaked value, accepted replay, executed payload, or unauthorized result).
5. Contrast with the secure path. Run the module’s secure/hardened action (where provided) and confirm the same attack is rejected - this is what a correct implementation should do.
6. Map it back. Cross-reference the MASTG v2 test(s) MASTG-TEST-0027, MASTG-TEST-0026 for the canonical procedure and remediation.

Tools (optional)

• garak
• promptfoo
• frida

Android

• adb

iOS

• xcrun simctl openurl

garak / promptfoo / MCP scanner test the LLM or MCP endpoint behind the app, not the app binary. Point them at the backend model API (find it with mitmproxy / Burp Suite) or a local on-device model server; the in-app demo already exercises the same prompt path.

Attack inputs

Payloads/artifacts you author for the on-device attack. The demo already ships and simulates these in-process (e.g. the malicious companion app / crafted intent is emulated inside the screen), so you only need to craft them to reproduce the exploit on a real device:

• crafted prompt

Real-world references

Concrete public disclosures that match this vulnerability class:

• CVE-2025-5020 (Firefox iOS URL scheme spoofing)
• CVE-2026-35394 (Mobile MCP arbitrary Android intent via mobile_open_url)