e2e-test
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external inputs such as test recordings, user-supplied prompts, and git diffs to generate or modify E2E tests. This data ingestion creates an attack surface for indirect prompt injection.\n
- Ingestion points: Workflow modes ingest recording JSON, user prompts, and git diffs (SKILL.md).\n
- Boundary markers: The skill references external quality protocols (e.g.,
.claude/skills/shared/e2e-quality-protocol.md) for defining test gates.\n - Capability inventory: The skill executes project-configured shell commands, performs file-system operations (writing reports), and spawns specialized sub-agents.\n
- Sanitization: Instructions mandate redacting sensitive values and require adherence to fidelity gates.\n- [COMMAND_EXECUTION]: The skill identifies and executes end-to-end test commands specified within project-level configuration files (e.g.,
docs/project-config.json). It provides explicit constraints to the agent to use only verified, copy-ready commands.\n- [DYNAMIC_EXECUTION]: The skill manages complex workflows including system startup, system exercise, and test execution. It delegates specialized generation tasks and visual baseline updates to a dedicatede2e-runnersub-agent based on the detected project framework.
Audit Metadata