The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill instructs the user to execute echo 0 | sudo tee /proc/sys/kernel/yama/ptrace_scope to disable kernel-level ptrace protections. This modification allows the debugger to attach to arbitrary processes, which bypasses default security configurations and increases the overall system attack surface.
[DYNAMIC_EXECUTION]: The skill utilizes tools (pdb, debugpy, remote-pdb) that facilitate arbitrary Python code execution at runtime. Specifically, the !stmt and interact commands in pdb allow for executing any Python statement, and debugpy supports attaching to and executing code within running processes via the --pid flag. Additionally, the skill includes a recipe for generating a custom DAP client script in /tmp/dap_client.py for dynamic execution.
[COMMAND_EXECUTION]: The instructions include the use of sudo and shell commands for configuration changes and installing debugging dependencies via the package manager.
[INDIRECT_PROMPT_INJECTION]: The skill presents an indirect prompt injection surface when debugging untrusted code or test suites.
Ingestion points: Processes external code and data through pdb, pytest, and debugpy as described in SKILL.md.
Boundary markers: No specific delimiters or instructions to ignore embedded instructions in the debugged source code are provided.
Capability inventory: Full arbitrary Python execution via pdb (!stmt), interact mode, and debugpy process injection.
Sanitization: No sanitization or validation of the input code or data processed during a debugging session is implemented.

python-debugpy