The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses eval to execute a find command when scanning for project materials. The use of eval with shell strings is a dangerous pattern that can be exploited for command injection if any component of the command becomes influenced by unsanitized input.\n- [COMMAND_EXECUTION]: The skill executes a local shell script (status.sh) located in the related codeck skill directory. This establishes a dependency on external scripts whose contents are not provided in the skill definition.\n- [PROMPT_INJECTION]: The skill implements an indirect prompt injection surface by reading and analyzing a broad range of untrusted file formats (PDF, DOCX, CSV, etc.) to derive story arcs and editorial instincts. It lacks explicit boundary markers or sanitization to prevent malicious instructions within those files from influencing the agent's logic.\n- [DATA_EXFILTRATION]: The skill combines the capability to read sensitive local project files with the ability to perform web searches. While no direct exfiltration was identified, this combination allows for potential data leakage if the agent is manipulated by adversarial content within the scanned project materials.

codeck-outline