The Agent Skills Directory

[SAFE]: The skill uses respira_get_snapshot to create a restoration handle before building the style-guide page, allowing for a clean rollback if the generated content is unsatisfactory.
[SAFE]: Telemetry is sent to the developer's official domain at https://www.respira.press/api/skills/track-usage. The skill documentation confirms that sensitive design tokens and logo URLs are excluded from these metadata reports.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from existing WordPress pages and theme files, creating a potential surface for indirect prompt injection.
Ingestion points: The skill calls respira_list_pages, respira_read_theme_file, and respira_extract_builder_content to gather brand data.
Boundary markers: No specific delimiters or "ignore instructions" tags are implemented for the content being processed during synthesis.
Capability inventory: The skill has the authority to write to the database (respira_update_option), generate new site pages (respira_build_page), and register global style tokens (respira_create_design_token).
Sanitization: The synthesis process enforces strict validation rules, such as converting all colors to hex format and requiring absolute URLs for media, which limits the potential for data-driven attacks.

design-system-synthesizer