The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the gdb-cli package from PyPI and directly from a GitHub repository (https://github.com/Cerdore/gdb-cli.git). These are third-party sources not associated with well-known trusted vendors.
[REMOTE_CODE_EXECUTION]: By installing and then running commands from the gdb-cli package, the skill executes code retrieved from external, community-maintained sources. This includes both Python code and the underlying GDB extensions provided by the package.
[COMMAND_EXECUTION]: The skill makes extensive use of shell commands to interact with the system, including pip install, gdb, and various gdb-cli subcommands like attach, load, and eval-cmd. These commands are used to inspect live process memory and core dumps.
[DATA_EXFILTRATION]: The skill provides tools for deep memory inspection (gdb-cli memory) and variable evaluation (gdb-cli eval-cmd). While intended for debugging, these capabilities allow for the extraction of sensitive information from process memory or core dumps if used maliciously or on unauthorized data.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) as it ingests untrusted data from external sources such as core dumps, binary files, and source code. Maliciously crafted data in these files could potentially influence agent behavior when the agent processes the output of the debugging commands. The skill lacks explicit boundary markers or sanitization steps for this external data.

gdb-cli