The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted conversation data to generate visual artifacts, creating an attack surface for indirect injection.
Ingestion points: The skill processes the "current topic of conversation" as the primary input for generating visuals, as noted in SKILL.md.
Boundary markers: The instructions lack explicit delimiters or warnings to ignore instructions embedded within the conversation data being visualized.
Capability inventory: The skill uses Bash(open ...) to open files generated from conversation data (SKILL.md).
Sanitization: There are no instructions provided for sanitizing or escaping user-controlled content before it is written to the HTML artifact.
[DYNAMIC_EXECUTION]: The skill instructions direct the agent to generate focused HTML artifacts (diagrams, infographics, or slide decks) at runtime and then immediately open them using a shell command.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute an open command on a file path containing a variable {description} derived from user-influenced content.

show-me