multi-user-workflow-to-playwright
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes workflow markdown files to generate sub-agent prompts and test code, creating a surface for indirect prompt injection. Ingestion points: The workflow file /workflows/multi-user-workflows.md serves as the data source. Boundary markers: The agent parses the markdown into a structured data model before using it in prompts. Capability inventory: Generated tests use standard Playwright APIs; exploration sub-agents are limited to read-only file system operations. Sanitization: The skill relies on structured parsing to isolate workflow content from instructional logic.
- [SAFE]: Credential management in generated code uses environment variables and GitHub secrets, avoiding exposure of sensitive information.
- [SAFE]: The installation phase for the generated project uses the --ignore-scripts flag, mitigating risks from malicious package lifecycle scripts.
- [SAFE]: The skill generates projects within a scoped directory (e2e/multi-user/) and does not perform unauthorized system modifications or persistence attempts.
Audit Metadata