python-debugpy
Warn
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: MEDIUMPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [PRIVILEGE_ESCALATION]: The skill provides instructions to execute a sudo command to modify system-level kernel security settings by writing to '/proc/sys/kernel/yama/ptrace_scope'. This modification reduces the operating system's security by allowing processes to trace and inspect the memory of other processes, a capability required for attaching debuggers to running processes but which increases the system's attack surface.\n- [COMMAND_EXECUTION]: The instructions include numerous shell commands for package management, executing test scripts like 'scripts/run_tests.sh', and using network utilities such as 'nc' to establish remote debugging sessions.\n- [EXTERNAL_DOWNLOADS]: The skill recommends installing external Python packages including 'debugpy' and 'remote-pdb' using pip. While 'debugpy' is a well-known package from a trusted provider, the installation of unversioned third-party packages at runtime presents a dependency risk.\n- [INDIRECT_PROMPT_INJECTION]: The skill exposes a large attack surface by facilitating an interactive Python REPL (pdb) and a Debug Adapter Protocol (DAP) interface which are designed to ingest and execute commands.\n
- Ingestion points: Debugger REPL prompts and network sockets (DAP/remote-pdb).\n
- Boundary markers: None; the environment is designed for direct command entry.\n
- Capability inventory: Full ability to execute arbitrary Python code, access the file system, and perform network operations.\n
- Sanitization: None; the tool is intended to provide unrestricted access to process state for debugging purposes.\n- [DYNAMIC_EXECUTION]: The primary function of the skill is to enable the dynamic execution of Python code for debugging and state inspection, incorporating the use of debugger commands that effectively wrap 'eval' and 'exec' functionality.
Audit Metadata