ln-621-security-boundary-auditor
Pass
Audited by Gen Agent Trust Hub on May 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a security auditing workflow intended to identify vulnerabilities in source code. It follows a two-layer detection process involving candidate scanning and context verification.
- [SAFE]: File access and analysis are performed using standard local tools (Read, Grep, Glob, Bash) and specialized MCP tools (hex-graph, hex-line) provided by the platform vendor. No remote code execution or unauthorized network exfiltration patterns were detected.
- [SAFE]: No evidence of prompt injection, obfuscation, or persistence mechanisms was found in the skill's instructions or metadata.
- [SAFE]: Indirect Prompt Injection Surface: The skill's primary function is to ingest and analyze untrusted codebase content. This constitutes a data ingestion surface, but it is inherent to the auditing purpose and managed by the skill's reliance on structured regex patterns and analytical verification rather than uncontrolled execution of input data.
Audit Metadata