The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input via the design brief ($ARGUMENTS) and propagates it to five parallel sub-agents without sufficient isolation or sanitization.
Ingestion points: SKILL.md (The brief is parsed and included in the Design Problem Statement).
Boundary markers: Absent; the instructions do not use delimiters or explicit 'ignore instructions' warnings when passing the brief to sub-agents.
Capability inventory: Sub-agents have access to powerful design tools including batch_design (file modification), Task (spawning sub-tasks), and get_screenshot.
Sanitization: The skill does not perform any validation or filtering on the brief content before processing.
[EXTERNAL_DOWNLOADS]: The feedback section provides a command (npx --yes hivenet@latest) that downloads the hivenet package from the npm registry at runtime. This tool is part of the author's (zvadaadam) feedback infrastructure for the skill ecosystem.
[COMMAND_EXECUTION]: The skill provides a pre-formatted shell command for the user to submit feedback via the hivenet CLI tool.

design-roundtable