folddevtools-android-webview-debugger
Fail
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: CRITICALEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Automated scanning flagged the URL
https://another-site.comas malicious/blacklisted. This URL is included in the skill's Kotlin code implementation examples (SKILL.md). - [EXTERNAL_DOWNLOADS]: The skill instructs users to download and install a binary APK from an unverified third-party GitHub repository (
achyuki/FoldDevtools). - [COMMAND_EXECUTION]: Provides instructions for executing multiple ADB commands that interact with the Android system, including socket discovery (
adb shell cat /proc/net/unix), port forwarding (adb forward), and activity management (adb shell am start). - [PRIVILEGE_ESCALATION]: Recommends the use of root access and XPosed/LSPosed modules to enable system-wide WebView debugging, which involves significant modification of the device's security model.
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection via system command output processing.
- Ingestion points: The agent is instructed to read and process output from system commands like
adb shell cat /proc/net/unixandadb shell ps(SKILL.md). - Boundary markers: The skill lacks delimiters or explicit instructions to prevent the agent from misinterpreting malicious socket names or process strings as control instructions.
- Capability inventory: The skill uses data gathered from the system to construct and execute sensitive shell commands such as
adb forwardandadb shell am start(SKILL.md). - Sanitization: There is no mention of validation, filtering, or escaping of system-derived strings before they are incorporated into subsequent execution steps.
Recommendations
- AI detected serious security threats
- Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata