superclaude
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill formalizes a hierarchy of instructions stored in CLAUDE.md files that the agent is prompted to load and obey. This creates an attack surface where a contributor could influence agent behavior by modifying these files.
- Ingestion points: The agent is instructed to read instructions from root and subdirectory CLAUDE.md files across the project (SKILL.md).
- Boundary markers: The recommended configuration templates lack explicit delimiters or instructions to treat codebase content as untrusted when processing these configuration files.
- Capability inventory: The harness enables the agent to execute shell commands (e.g., npm test, npm run lint) and invoke subagents, which could be abused if the agent is misled by injected instructions (SKILL.md).
- Sanitization: No sanitization or verification procedures are defined for the instructions parsed from the codebase hierarchy.
Audit Metadata