The Agent Skills Directory

[SAFE]: The skill performs its stated function of design analysis using appropriate tools. It retrieves data via a dedicated MCP server (Stitch) and generates documentation files. No unauthorized network activity or credential exposure was detected.
[PROMPT_INJECTION]: The skill is designed to ingest and process external HTML/CSS code, which creates a potential surface for indirect prompt injection.
Ingestion points: HTML source code is downloaded via web_fetch from URLs provided by the MCP server (documented in SKILL.md).
Boundary markers: There are no explicit markers or instructions to the agent to disregard instructions potentially hidden within the processed HTML comments or strings.
Capability inventory: The skill utilizes web_fetch for network access and Write for file generation.
Sanitization: The skill extracts design tokens through parsing but does not include specific sanitization logic to strip executable content or instructions from the input data.

design-md