The Agent Skills Directory

[COMMAND_EXECUTION]: Several scripts, including fill_template.py, layout_gate.py, and ole2_to_docx.py, utilize the subprocess.run function to execute shell commands. These calls are primarily used to invoke the soffice (LibreOffice) binary for converting documents between formats (e.g., Docx to PDF). While these operations are essential for the skill's primary purpose of generating valid legal documents, they involve the execution of external binaries based on local environment configurations and file paths.
[INDIRECT_PROMPT_INJECTION]: The skill implements a workflow that uses an AI agent to perform semantic extraction of elements from user-provided Markdown or Docx files, as described in SKILL.md and references/extraction-prompt-template.md. This ingestion of untrusted data creates a surface for indirect prompt injection.
Ingestion points: The AI agent reads user-supplied legal complaints or dialogue to produce structured JSON via the instructions in references/extraction-prompt-template.md.
Boundary markers: The prompt template does not explicitly include delimiters or instructions to ignore potential commands embedded within the user's source text.
Capability inventory: The skill possesses the ability to write to the file system and execute shell commands (via the conversion logic in fill_template.py).
Sanitization: The risk is mitigated by the fact that the output of the LLM is restricted to a structured JSON format (elements.json), which is then processed by deterministic Python code using lxml to perform string replacements in XML, preventing the user data from being executed as code.

elements-complaint-generator