selenium-cucumber-expert

Pass

Audited by Gen Agent Trust Hub on Sep 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill instructions and provided code samples focus on established software engineering principles for test automation. The content correctly handles sensitive data by advising the use of environment variables and CI/CD secrets rather than hardcoding credentials.
  • [COMMAND_EXECUTION]: The skill provides a Node.js utility scripts/scaffold-selenium-bdd.mjs to automate the creation of project directories and boilerplate files. This script performs local file system operations that are transparent and consistent with its stated purpose of project scaffolding.
  • [EXTERNAL_DOWNLOADS]: Documentation and configuration files reference standard package registries like Maven Central and infrastructure tools such as Docker Hub and GitHub Actions. These downloads target official repositories of well-known services for dependency management and environment setup.
  • [INDIRECT_PROMPT_INJECTION]: As a specialized code-generation assistant, the skill ingests user requirements to produce Selenium test scripts. It manages the associated risks by enforcing strict architectural patterns (Page Object Model and Screenplay) and providing rigid templates for code output.
  • Ingestion points: User requests for specific Gherkin features or test scenario implementations in SKILL.md.
  • Boundary markers: The skill emphasizes the use of Gherkin syntax and Java class structures as clear boundaries for generated content.
  • Capability inventory: Local file system writes via the scaffolding script and browser automation via Selenium WebDriver 4.
  • Sanitization: The scaffolding logic uses predefined templates with simple string interpolation for project and package names.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 27, 2026, 10:39 AM