The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from HTML and CSS files within the docs/product/demo/ directory to extract visual component definitions. Maliciously crafted content within these source documents could potentially influence the documentation generated in COMPONENTS.md or the mapping rules in component-map-rule.md, which may be utilized by other agents in the development pipeline.
Ingestion points: The skill reads HTML and CSS files from docs/product/demo/ during the component extraction phase (SKILL.md Step 1 and Step 4).
Boundary markers: The subagent instructions in templates/component-prompt.md do not include explicit delimiters or warnings to differentiate between the data being processed and the extraction instructions.
Capability inventory: The skill has the capability to create directories and write files to the docs/specs/design/ path (SKILL.md Step 3 and Step 4).
Sanitization: There is no explicit sanitization or filtering of the source HTML/CSS content described in the skill's logic beyond the model's inherent extraction capabilities.

specs-components