security-scan
Warn
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to run
npx @claude-flow/cli@latestto perform various scan operations. This dynamically fetches and executes an unpinned package from the public npm registry at runtime without version lock or checksum verification. - [INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection attacks.
- Ingestion points: Ingests untrusted external codebase content and PR data when conducting security audits.
- Boundary markers: Completely absent; there are no instructions or delimiters designed to isolate natural language content embedded within code files from the agent's instructions.
- Capability inventory: Employs powerful execution tools including
Bash,Read,Grep, and core MCP storage/hook plugins. - Sanitization: No sanitization, escaping, or strict validation is applied to code files prior to processing.
Audit Metadata