rust-code-navigator
Fail
Audited by Gen Agent Trust Hub on Feb 15, 2026
Risk Level: HIGHPROMPT_INJECTION
Full Analysis
- Obfuscation (HIGH): Detected multiple zero-width space characters (U+200B) hidden before code block markers. These characters are typically used to bypass automated security filters or hide malicious strings.
- Indirect Prompt Injection (HIGH): The skill is highly vulnerable because it reads and interprets external source code without sanitization. 1. Ingestion points: File content is ingested via LSP operations and the Read tool from files such as src/main.rs. 2. Boundary markers: Absent; no delimiters are used to separate code from instructions. 3. Capability inventory: The tool has Read and Glob permissions, which can be misused to access sensitive project data. 4. Sanitization: None; external code content is passed directly into the agent prompt context.
- Automated Scan Alert (MEDIUM): An external scanner (URLite) flagged the file path main.rs as malicious, indicating that the skill example code or logic may trigger security signatures related to known threats.
Recommendations
- AI detected serious security threats
- Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata