The Agent Skills Directory

[SAFE]: The skill's primary operations—creating, auditing, and fixing redirects—are well-defined and include functional safety checks such as status code verification and loop detection.- [DATA_EXFILTRATION]: The skill accesses application-specific configuration files located in ~/.claude-marketing/ to retrieve brand context, profiles, and compliance rules. These file accesses are scoped to the tool's marketing-specific purpose and do not target generic sensitive system credentials.- [COMMAND_EXECUTION]: Facilitates the modification of web server configurations (e.g., .htaccess) and CMS-specific redirect rules via connected MCP servers. The instructions specify these as 'last resort' operations and focus on legitimate administrative tasks.- [PROMPT_INJECTION]: Ingests potentially untrusted data from source URLs and bulk-import sources like CSV or Google Sheets. Ingestion points: User-supplied URLs and import files; Boundary markers: Absent; Capability inventory: CMS API writes, .htaccess modification, and logging via seo-executor.py; Sanitization: Implements HTTP status code checks (200 OK) and path validation to ensure targets are valid before processing.- [REMOTE_CODE_EXECUTION]: References a local script named seo-executor.py used to log execution metadata and maintain an audit trail. This is a standard pattern for extending agent functionality with task-specific automation scripts.

redirect-manager