test-integration
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The integration runner installs external dependencies documented in markdown files (such as NuGet packages, Python libraries, and Node.js modules) into disposable environments to validate example code.
- [DYNAMIC_EXECUTION]: The skill describes a harness that extracts and executes code snippets directly from documentation files across multiple languages including Python, JavaScript/TypeScript, Go, and .NET.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it processes and executes content from documentation files that could theoretically be modified to include malicious instructions.
- Ingestion points: Documentation files (
.mdx) listed inintegrations.jsonare parsed for code extraction. - Boundary markers: The runner relies on delimited sections for installation, setup, and execution to isolate code from prose.
- Capability inventory: The system can install arbitrary packages and execute code in various runtimes.
- Sanitization: The skill assumes the documentation files within the repository are trusted and does not perform additional sanitization of the extracted code before execution.
Audit Metadata