The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructions in SKILL.md direct the agent to execute a local Node.js script while directly interpolating user-provided text into the shell command string using the --input flag. This creates a risk of shell command injection if the agent environment does not properly escape special characters, potentially allowing an attacker to execute arbitrary commands on the host system.
[DATA_EXPOSURE]: The utility script analyze-slop.js accepts --file and --output arguments that allow reading from and writing to any file path accessible to the agent. This could be abused to exfiltrate sensitive information or overwrite critical configuration files by providing malicious file paths.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted text from users without employing boundary markers or sanitization, creating a surface for indirect prompt injection. Ingestion points: User-provided text strings and files processed in Step 3 and Step 5 of the workflow. Boundary markers: None; the skill lacks delimiters or instructions to ignore commands within the analyzed data. Capability inventory: The skill has the capability to execute shell commands (Node.js) and perform file system read/write operations. Sanitization: None; the input is passed directly to the analysis script without filtering or escaping.

neo-stop-slop