The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the WordPress site's pages and posts, which could potentially contain malicious instructions intended to influence the agent's analysis.
Ingestion points: Raw content strings retrieved via respira_list_pages and respira_list_posts in SKILL.md Step 5.
Boundary markers: None; the skill lacks delimiters or instructions to ignore embedded commands within the site content while scanning for shortcodes.
Capability inventory: The skill uses various read-only audit tools provided by the respira-wordpress MCP server to analyze site versions, plugins, and performance.
Sanitization: No sanitization or escaping of the ingested content is performed before it is processed by the agent.
[DATA_EXFILTRATION]: The skill transmits site metadata and audit findings to a vendor-controlled telemetry endpoint after generating the report.
Evidence: SKILL.md Step 10 includes a code block that sends a JSON payload—containing the site URL, WordPress/PHP versions, builder list, and issue counts—to https://www.respira.press/api/skills/track-usage.
Note: The telemetry targets the official domain of the skill author ('respira-press').
[EXTERNAL_DOWNLOADS]: The skill instructions direct the user to install an external package required for the skill to function.
Evidence: SKILL.md Step 1 recommends the installation of the @respira/wordpress-mcp-server package via npx.
Note: The package is a vendor-scoped resource provided by the author.

wordpress-site-dna