Untrusted Mobile Input -> LLM Prompt (deep link / clipboard / QR)

← AI + Mobile (LLM x IPC / WebView)

untrusted_mobile_input_to_llm   MEDIUM

CategoryAI + Mobile (LLM x IPC / WebView)
OWASP Mobile (2024)M4
MASVSMASVS-PLATFORM-3MASVS-CODE-4
MASWEMASWE-0050
CWECWE-77CWE-20
PlatformAndroidiOS

Description

Content arriving over a mobile trust boundary (deep-link param, clipboard, scanned QR, notification) is concatenated straight into the assistant’s prompt, turning classic mobile IPC into a prompt-injection delivery channel (Monica CVE-2024-48142 class).

How it works

Content arriving over a mobile trust boundary (a deep-link query param, the clipboard, a scanned QR code, a notification) is concatenated straight into the assistant prompt with no separation from the trusted system instruction. Because that IPC channel is attacker-controllable, it becomes a prompt-injection delivery channel: instructions smuggled through the deep link override the system prompt and drive an unconfirmed tool call that exfiltrates the system-prompt secret (the Monica CVE-2024-48142 class). The secure path treats the mobile input as inert, quoted data and screens it with an injection guard, so nothing fires.

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 · Android: adb · iOS: xcrun simctl openurl.
  2. Locate the target. From the home index, open Untrusted Mobile Input -> LLM Prompt (deep link / clipboard / QR) (untrusted_mobile_input_to_llm). 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.

Tools (optional)

Android

iOS

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
  • crafted deep link
  • crafted QR

Real-world references

Concrete public disclosures that match this vulnerability class: