The Agent Skills Directory

[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill instructs the installation of the gdb-cli package from PyPI or directly from the author's GitHub repository (https://github.com/Cerdore/gdb-cli.git). As this resource belongs to the skill's author, it represents the primary functionality of the tool.
[DYNAMIC_EXECUTION]: The skill uses the gdb-cli eval-cmd function to evaluate GDB expressions. This allows for dynamic inspection of the debuggee's state, which is a standard feature of debugging tools but involves runtime evaluation of logic.
[DATA_EXPOSURE_AND_EXFILTRATION]: The skill provides commands to read sensitive runtime information, including process memory (gdb-cli memory), register values (gdb-cli registers), and local variables (gdb-cli locals-cmd). These operations are inherent to the skill's purpose as a debugger and are performed locally within the session.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes untrusted data from external sources.
Ingestion points: The agent reads data from core dumps (gdb-cli load), process memory (gdb-cli memory), and source code files during correlation.
Boundary markers: None are present in the provided instructions.
Capability inventory: The skill can execute shell commands via gdb-cli and attach to system processes.
Sanitization: There is no explicit evidence of sanitization or filtering of the content read from the debugged processes or files.

gdb-cli