The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill explicitly facilitates Remote Code Execution (RCE) by providing guidance and targets for unauthenticated file upload vulnerabilities (e.g., CVE-2024-2534 in Slider Revolution and CVE-2023-6853 in ElementsKit).
[COMMAND_EXECUTION]: The skill contains several shell scripts that use curl, grep, and sed to perform automated scanning of remote targets. It also attempts to create directories and write output files to /root/output/plugins/, which is a restricted system path requiring elevated privileges.
[DATA_EXFILTRATION]: The skill is designed to identify and extract sensitive information from third-party sites, including WooCommerce customer data, author emails via Yoast SEO sitemaps, and Gravity Forms entries.
[CREDENTIALS_UNSAFE]: The skill describes methods for authentication bypass and privilege escalation, such as manipulating LiteSpeed Cache tokens (CVE-2024-50550) and exploiting Gravity Forms (CVE-2024-6115).
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface through the ingestion of untrusted external data.
Ingestion points: Processes raw HTML and JSON data from external target servers (e.g., PAGE=$(curl ...) and readme=$(curl ...) in Steps 2 and 3).
Boundary markers: No boundary markers or instructions to ignore embedded content are present.
Capability inventory: Includes network access via curl and file system writes via mkdir and cp.
Sanitization: No sanitization or validation of the ingested external content is performed before processing.

wordpress-plugin-hunt