The Agent Skills Directory

[SAFE]: No malicious patterns, such as hardcoded credentials, data exfiltration, or persistence mechanisms, were found. The skill operates within its intended scope of generating text-based prompt templates.
[COMMAND_EXECUTION]: The skill instructs the agent to use standard shell commands (cat, head, grep) for internal operations. cat is used to read reference files provided in the skill package, while head and grep are used in Step 7 to validate the syntax and structure of the generated template files.
[PROMPT_INJECTION]: The skill processes user-provided task descriptions and requirements to generate templates, representing an indirect prompt injection surface. However, the risk is negligible as the output consists of static markdown files intended for human-in-the-loop usage. Evidence chain: 1. Ingestion points: User descriptions and task analysis in Steps 2-5; 2. Boundary markers: Markdown sections and headers; 3. Capability inventory: File write and standard shell command execution; 4. Sanitization: Syntax validation steps included in the skill workflow.

generate-prompt