The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a variety of shell commands including grep, find, ls, cat, and jq to inspect the local filesystem and configuration files for security vulnerabilities.
[DATA_EXPOSURE]: The skill instructions direct the agent to read sensitive locations such as ~/.claude.json, ~/.ssh/, and .env files to check for exposed credentials or private keys. This behavior is consistent with the skill's stated purpose of performing a security audit.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a parser for untrusted data (the contents of other skills and configuration files). It identifies vulnerability surfaces by scanning for prompt injection keywords like 'ignore', 'override', and 'disregard' in memory and config files. While it lacks hard boundaries when interpolating this data into its report, the risk is inherent to its function as a security scanner.
[REMOTE_CODE_EXECUTION]: The skill explicitly audits other files for remote code execution patterns, such as curl | bash or wget | sh, effectively acting as a defensive mechanism rather than a threat vector.

security-check