gin-api
Warn
Audited by Gen Agent Trust Hub on Mar 1, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Path traversal vulnerability in routing documentation. In 'references/routing.md', the 'UploadDocuments' code example uses 'file.Filename' directly to create a destination path: 'fmt.Sprintf("uploads/docs/%s", file.Filename)'. Since 'file.Filename' is provided by the client and not sanitized, an attacker could supply a filename containing directory traversal sequences (e.g., '../../etc/passwd') to write files to arbitrary locations on the host system.
- [PROMPT_INJECTION]: Indirect prompt injection attack surface.
- Ingestion points: Untrusted data enters the agent context via 'c.ShouldBindJSON', 'c.ShouldBindQuery', and 'c.FormFile' in 'SKILL.md' and 'references/routing.md'.
- Boundary markers: The skill implements request binding with validation tags (e.g., 'binding:"required,email"') but lacks explicit delimiters or instructions to prevent the processing of malicious commands embedded in user-supplied strings.
- Capability inventory: The skill facilitates file system writes ('c.SaveUploadedFile') and service-layer business logic execution.
- Sanitization: The skill lacks sanitization of external content before its use in sensitive operations, specifically file path construction.
Audit Metadata