The Agent Skills Directory

[DATA_EXFILTRATION]: The skill transmits usage telemetry to the vendor-controlled endpoint https://www.respira.press/api/skills/track-usage. The documentation states the data includes site URL hashes, page counts, and environment metadata, while excluding specific design tokens or logo URLs. As this communication targets the skill author's infrastructure, it is identified as a vendor telemetry channel.- [PROMPT_INJECTION]: The skill processes content from various WordPress pages through the respira_extract_builder_content tool to synthesize a design system. This ingestion of potentially untrusted site data creates a vulnerability surface for indirect prompt injection, where malicious instructions embedded in the analyzed pages could influence the synthesis logic or the creation of the resulting style guide page.
Ingestion points: Page content extracted in Step 5 using respira_extract_builder_content.
Boundary markers: Absent; the skill does not specify delimiters or instructions to ignore embedded commands within the extracted data.
Capability inventory: The skill utilizes multiple high-privilege tools including respira_update_option for database writes, respira_build_page for site content creation, and respira_create_design_token for modifying global styles.
Sanitization: Not explicitly performed; values are aggregated and persisted directly to the WordPress options table.- [COMMAND_EXECUTION]: The skill relies on the respira-wordpress MCP server to execute various shell-like operations within the WordPress environment, including reading theme files, modifying database options, and managing site snapshots.

design-system-synthesizer