The Agent Skills Directory

[SAFE]: The skill performs legitimate metadata manipulation for the 1C:Enterprise platform, converting XML schemas into the platform-specific XDTO format.
[SAFE]: Includes a 'support-guard' mechanism in both PowerShell and Python scripts which acts as a safety feature to prevent the modification of locked vendor objects, protecting the user's configuration integrity.
[COMMAND_EXECUTION]: The skill utilizes local PowerShell and Python scripts to perform its tasks. These scripts are contained within the skill's own directory and operate on user-provided input paths.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it processes untrusted XML schemas (XSD).
Ingestion points: XSD file content provided via parameters is processed by scripts/xdto-compile.ps1 and scripts/xdto-compile.py.
Boundary markers: Absent; the tool returns warnings derived from the schema content directly to the agent.
Capability inventory: The scripts perform file writes to the project directory and modify the Configuration.xml file.
Sanitization: Standard XML parsers are used; no specific sanitization of schema content for natural language instructions is present.

xdto-compile