The Agent Skills Directory

[PROMPT_INJECTION]: The skill accepts user input via the $ARGUMENTS variable to define the advisor task. This creates a surface where user-provided text is interpolated into a prompt for a secondary agent. To mitigate risks, the skill includes a mandatory safety suffix that instructs the secondary agent not to perform file operations or execute code.
[DATA_EXFILTRATION]: The skill facilitates data access by instructing the agent to provide relevant file paths to the secondary advisor agent. This is a core feature for context-aware analysis but involves transferring file context to the secondary agent environment.
[COMMAND_EXECUTION]: The skill is design-oriented and does not contain any direct shell command execution or script invocation. All actions are performed through agent-to-agent communication and the paseo orchestration layer.

paseo-advisor