setup-claude
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill executes a legitimate administrative workflow for environment configuration. It uses platform-native tools for managing plugins and connectors as intended and provides users with options to skip or review steps.\n- [SAFE]: Reading user role and installed skill information from the system context is a standard feature for personalization and preventing redundant tasks. No sensitive file paths or credentials are accessed.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external tool outputs (marketplace and registry searches), which is a common attack surface.\n
- Ingestion points: Marketplace search results and connector registry data are processed in SKILL.md (Steps 2 and 3).\n
- Boundary markers: The instructions rely on tool-specific widgets to display data, reducing the likelihood of prompt injection from tool results.\n
- Capability inventory: The skill can search for and suggest plugins/connectors and invoke secondary setup skills.\n
- Sanitization: No specific sanitization is implemented, but the risk is aligned with the skill's primary setup function and is handled by the platform's tool architecture.
Audit Metadata