The Agent Skills Directory

[COMMAND_EXECUTION]: The skill contains numerous examples of curl commands used for security testing and vulnerability scanning against a target server defined by the $TARGET variable.
[REMOTE_CODE_EXECUTION]: While the automated scan flagged a potential 'remote code piped to node' pattern, manual review confirms this is a false positive. The command curl -sI https://$TARGET/ | grep -i "x-powered-by\|nodejs\|express" is a benign fingerprinting step that pipes the HTTP headers to grep for string matching, not to a node interpreter for execution.
[INDIRECT_PROMPT_INJECTION]: The skill includes payloads for testing Server-Side Template Injection (SSTI) and Prototype Pollution. These are documented as test cases for a security researcher to use against a target, rather than instructions meant to subvert the agent itself.
[DATA_EXFILTRATION]: The skill references /proc/self/environ as a target for Local File Inclusion (LFI) testing to identify environment variable leaks. This is a standard security testing practice and does not involve the skill itself exfiltrating data.

hunt-nodejs