cf-validate
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes PowerShell and Python scripts to perform validation logic on local files. The instructions provide examples of calling
powershell.exeto execute the provided scripts. - [DATA_EXPOSURE]: The validation process involves reading local 1C configuration files (e.g.,
Configuration.xml,Languages/*.xml,Ext/*.xml). These reads are limited to the path provided by the user and are used for structural analysis. - [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted XML configuration data. This presents a surface for indirect prompt injection if the agent interprets instructions within the data. However, the logic is constrained to specific XML tag and attribute validation using standard parsers (
lxml.etreeandSystem.Xml.XmlDocument), which reduces the risk. - Ingestion points: The skill reads
Configuration.xml, language files, and extension metadata files from the user-provided path. - Boundary markers: None present; the skill treats the XML content as data to be validated.
- Capability inventory: File system read operations and an optional file write operation via the
-OutFileparameter. - Sanitization: The scripts use established XML parsing libraries to handle the configuration data.
Audit Metadata