The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted XML configuration files (Rights.xml, Configuration.xml) provided by the user, which serves as a potential attack surface for indirect prompt injection.
Ingestion points: Reads XML content from file paths specified in the RightsPath argument and associated metadata files.
Boundary markers: No explicit instruction-ignoring delimiters are used when the agent processes the tool's validation output.
Capability inventory: The skill possesses file read and file write capabilities (via the OutFile parameter) in both PowerShell and Python scripts.
Sanitization: The scripts implement strict XML parsing and validation against internal dictionaries of known 1C object types and rights, which provides significant protection against malicious payload execution.
[COMMAND_EXECUTION]: The skill executes a PowerShell or Python script to perform the validation logic.
Evidence: The SKILL.md file contains a command definition that invokes powershell.exe with the -File parameter to run the bundled validation script.

role-validate