The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes external inputs and uses command feedback to modify its own instructions.
Ingestion points: Parameters such as recipient, file, and caption (SKILL.md), as well as command execution output used for the 'Self-Evolving Skill' logic.
Boundary markers: Absent. The skill does not define delimiters or provide instructions to ignore malicious content within the file paths or captions.
Capability inventory: The skill utilizes the Bash tool to execute bun scripts and implicitly encourages the agent to perform file modifications to 'fix' instructions.
Sanitization: Absent. Input variables are used directly in shell command examples, and there are no validation steps described for the 'self-evolution' updates.

send-media