The Agent Skills Directory

[COMMAND_EXECUTION]: The dbg launch and dbg attach commands provide the agent with the ability to execute arbitrary processes or attach to existing ones, which can be leveraged to run unauthorized code in the host environment.
[DATA_EXFILTRATION]: The dbg attach command allows connection to arbitrary WebSocket URLs, facilitating potential data transfer to external, non-whitelisted endpoints.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and displays untrusted data from external sources (process logs via dbg console and source code via dbg state or dbg source) without sanitization.
Ingestion points: dbg console, dbg state, dbg source, and dbg eval (reads runtime state and files).
Boundary markers: Absent; there are no instructions to the agent to ignore commands found within debugged output.
Capability inventory: dbg launch, dbg attach, dbg set, and dbg eval provide extensive system and process control.
Sanitization: Absent; the skill does not attempt to filter or escape the content retrieved from the debugged environment.

dbg