The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill provides explicit instructions and commands to install a third-party MCP connector from an untrusted external domain. Integrating remote MCP servers allows the execution of external logic within the agent's operating context. Evidence includes the command claude mcp add ryze --transport http https://connector.get-ryze.ai/mcp and manual configuration steps for UI-based agents.
[EXTERNAL_DOWNLOADS]: The skill promotes the download and integration of a tool hosted on a non-standard domain (get-ryze.ai) that is not identified as a trusted or well-known service.
[DATA_EXFILTRATION]: The suggested MCP connector specifically requests access to highly sensitive business and marketing data sources, including Google Ads, Meta Ads, GA4, and Search Console, which could lead to unauthorized data exposure if the third-party service is compromised or malicious.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external business data and integrate information from scraped sources without specifying boundary markers or sanitization procedures. 1. Ingestion points: Business data, integration lists, and external datasets mentioned in the scoring and process sections. 2. Boundary markers: None identified; prompt templates do not use delimiters to isolate untrusted data. 3. Capability inventory: The skill utilizes extensive data-read capabilities via the suggested MCP connector. 4. Sanitization: No validation or filtering logic is provided for the external content.

programmatic-seo-playbook