skd-decompile
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill performs localized data transformation on user-provided XML files without any network access or external connectivity, adhering to the principle of least privilege.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted XML input from external sources, which presents a surface for indirect prompt injection if the processed content is subsequently interpreted by the agent as instructions.
- Ingestion points: Untrusted data enters the agent context through the
TemplatePathargument in bothscripts/skd-decompile.ps1andscripts/skd-decompile.py. - Boundary markers: The skill does not implement explicit boundary markers or 'ignore embedded instructions' warnings in the generated JSON, Markdown, or SQL output files.
- Capability inventory: The skill leverages the
ReadandWritetools to process schemas and generate multiple output artifacts (JSON draft, warnings log, and SQL query files). - Sanitization: While the script validates that the input is a valid 1C DataCompositionSchema, it performs no sanitization of the textual expressions or queries contained within the XML before interpolating them into the output artifacts.
Audit Metadata