The Agent Skills Directory

[SAFE]: No malicious patterns, behaviors, or obfuscated content were detected in the skill assets or instructions. The skill acts as a technical resource for document format research.\n- [COMMAND_EXECUTION]: Provides usage examples for common shell utilities such as unzip and xmllint to inspect and format XML components of document archives.\n- [EXTERNAL_DOWNLOADS]: References the well-known xmlschema Python library for validating document parts against local schema files.\n- [PROMPT_INJECTION]: The skill has an indirect prompt injection attack surface as it processes raw document XML data.\n
Ingestion points: Raw document XML parts like word/document.xml or content.xml enter the agent context via inspection commands (SKILL.md).\n
Boundary markers: Absent; the provided command examples do not use delimiters to isolate document content.\n
Capability inventory: Uses unzip, xmllint, and Python's xmlschema to process external document files (SKILL.md).\n
Sanitization: Absent; no instructions are provided to sanitize or escape XML content before it is processed by the agent.

raw-document