The Agent Skills Directory

[SAFE]: The skill consists of natural language instructions for a design workflow and contains no malicious code or suspicious activity.
[NO_CODE]: The skill is purely informational markdown and does not include any executable scripts, shell commands, or external package installations.
[INDIRECT_PROMPT_INJECTION]: The skill identifies a surface where the agent processes external user responses to update design memory.
Ingestion points: User feedback responses collected during the taste check checkpoints in Step 4.
Boundary markers: The skill does not explicitly define delimiters for user input, relying on the agent's standard interaction layer.
Capability inventory: The skill facilitates updating state files such as design-state.md and recording signals in design-memory based on the feedback received.
Sanitization: The skill focuses on qualitative design feedback and does not specify technical sanitization for the inputs, which are treated as design preferences rather than executable content.

taste-feedback