jlink
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes external debugging utilities such as JLink.exe and arm-none-eabi-gdb via the subprocess module to interface with hardware probes.
- [DYNAMIC_EXECUTION]: The tool programmatically creates temporary .jlink command scripts from templates to execute sequences of probe instructions.
- [INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection. 1. Ingestion points: Output from J-Link/GDB processes and local binary files (ELF/HEX). 2. Boundary markers: The skill returns structured JSON to the agent and includes an 'operation_mode' setting to require user confirmation for destructive actions. 3. Capability inventory: The skill can execute specific local debugger binaries and read/write workspace configuration files. 4. Sanitization: Results are parsed using regular expressions and paths are normalized before use.
Audit Metadata