The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external Product Requirement Documents (PRDs) and requirement summaries to produce UI design specifications. This ingestion of untrusted external content creates a surface for indirect prompt injection.
Ingestion points: The skill accepts PRD file paths, pasted document content, and natural language requirements as input in the SKILL.md and examples/usage.md workflows.
Boundary markers: The instructions do not provide explicit boundary markers or delimiters (such as XML tags or unique separators) to help the agent distinguish between its instructions and potentially malicious data inside the PRDs.
Capability inventory: While the skill body contains a self-imposed rule to not call remote write tools, the skill's configuration does not explicitly restrict tool access in the YAML frontmatter, meaning the underlying agent environment retains access to all available tools.
Sanitization: The instructions include guidelines to use placeholder data for sensitive personal information, but they lack specific sanitization logic to detect or neutralize instructions embedded within the analyzed documents.
[EXTERNAL_DOWNLOADS]: The documentation references the installation of additional components from a remote source managed by the skill's author.
Evidence: The file examples/usage.md provides instructions for users to install another skill using the command npx skills add full-stack-skills/stitch-skills --skill stitch-ui-designer.
Note: This resource belongs to the skill's author (full-stack-skills) and represents a standard installation procedure for the vendor's ecosystem.

stitch-ui-design-spec-generator