The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill fetches data from remote WordPress target sites and pipes the output directly to a python3 interpreter using the -c flag for parsing. This pattern of processing untrusted remote data through a script interpreter is a high-risk practice that can lead to remote code execution or interpreter exploitation if the target server is compromised or provides malicious data. \n- [COMMAND_EXECUTION]: Several shell scripts within the skill (wp-plugin-scan.sh, cve-matcher.sh, wp-bulk-vuln-check.sh) process variables derived from external network responses, such as plugin versions and domain names. These variables are interpolated into shell commands (including curl, printf, and case statements) without rigorous validation or sanitization, creating a surface for command injection. \n- [EXTERNAL_DOWNLOADS]: The skill is designed for mass network scanning and performs unauthenticated requests to arbitrary external domains. It utilizes the -k (or --insecure) flag with curl, which explicitly disables SSL/TLS certificate verification. This practice bypasses critical security protocols and exposes the agent's communications to man-in-the-middle attacks. \n- [PROMPT_INJECTION]: The skill exhibits a significant surface for indirect prompt injection. \n
Ingestion points: Untrusted data enters the environment from remote WordPress targets via curl operations throughout the skill. \n
Boundary markers: There are no delimited boundaries or explicit instructions to ignore embedded commands within the processed data. \n
Capability inventory: The skill utilizes shell execution, curl for network access, and python3 for data processing. \n
Sanitization: The skill relies on simple text filtering (e.g., grep, head) which is insufficient to sanitize or validate the integrity of the data before it influences the agent's logic or is used in subsequent commands.

wp-plugin-automation