The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/gdb_debugger.py executes system binaries like openocd and arm-none-eabi-gdb using Python's subprocess module. Commands are passed as lists to avoid shell injection vulnerabilities, following security best practices for process automation.
[DYNAMIC_EXECUTION]: The skill dynamically generates GDB initialization scripts (specifically in scripts/gdb_debugger.py's generate_gdb_script function) to automate complex debugging tasks like capturing crash context. These scripts are written to temporary files and executed in batch mode.
[INDIRECT_PROMPT_INJECTION]: The skill processes data returned from hardware debuggers, including register values and memory contents, and incorporates them into the agent's context. This creates a surface where malformed target state could potentially influence the agent.
Ingestion points: Parsing of tool output in scripts/gdb_debugger.py (lines 310-327).
Boundary markers: None explicitly present in the script's parsing logic.
Capability inventory: The skill has the ability to write temporary files and execute shell commands via subprocess across scripts/gdb_debugger.py.
Sanitization: The script performs basic filtering, checking for specific keywords (like 'sp', 'pc', 'lr') and formatting results before returning them to the agent.

debug-gdb-openocd