adb-device-unlock
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests a user-supplied PIN and interpolates it directly into a shell command string for execution on an external device.
- Ingestion points: The
<USER_SUPPLIED_PIN>placeholder inSKILL.mdrepresents untrusted data entering the execution context. - Capability inventory: The skill utilizes
adb shellto execute commands on the connected Android phone. - Sanitization: There is no evidence of sanitization or character escaping for the PIN input. A PIN containing shell metacharacters (e.g.,
;,&,|) could lead to command injection on the Android device's shell. - Boundary markers: The instructions do not specify boundary markers or validation logic to ensure the input is strictly numeric or within expected bounds.
- [COMMAND_EXECUTION]: The skill directs the agent to execute shell commands via the Android Debug Bridge (ADB) to interact with hardware.
- Evidence: Multiple instances of
adb -s "$SERIAL" shell ...are provided as primary functional instructions inSKILL.md.
Audit Metadata