interactive-gdb-debugging
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the execution of shell commands to manage tmux sessions, launch debuggers like GDB and GDB-multiarch, and control QEMU instances. These operations are essential to the skill's primary function of interactive debugging.
- [INDIRECT_PROMPT_INJECTION]: The workflow relies on the agent capturing and interpreting output from the GDB debugger, which may include data from untrusted binaries or remote targets, creating a surface for indirect prompt injection.
- Ingestion points: Untrusted data enters the agent context via
tmux capture-paneandtmux pipe-panecalls inSKILL.md. - Boundary markers: The instructions suggest waiting for the
(gdb)prompt as a synchronization point, but do not provide specific markers to isolate potentially malicious data within the output. - Capability inventory: The skill has the capability to execute further shell commands via
tmux send-keysandtmux new-sessionacross all stages of the workflow. - Sanitization: There is no evidence of sanitization, filtering, or escaping of the captured process output before it is evaluated by the agent.
Audit Metadata