package-cascade
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from the Datex marketplace and generated cascade plans which serve as an ingestion point for external content.
- Ingestion points: Data is ingested from
dxs marketplace searchresults and theplan.jsonfile generated during the planning phase inreferences/cascade-workflow.md. - Boundary markers: Instructions require parsing structured JSON output and utilize clear phase-based workflow boundaries.
- Capability inventory: The skill can execute repository operations, create feature branches, and publish packages using
dxs source cascade run. - Sanitization: The skill mandates 'Safety gates' including the resolution of display names to unique identifiers and explicit user confirmation of the target tenant and version before any execution.
- [COMMAND_EXECUTION]: The skill generates and executes shell commands for the
dxsCLI to manage the dependency graph. - Evidence: Executes commands such as
dxs source cascade plan,dxs source cascade run, anddxs auth loginas described inSKILL.mdandreferences/cascade-workflow.md. - Context: These operations are part of the intended primary purpose of the skill for developer environment management within the vendor's ecosystem.
Audit Metadata