frontend-from-backend
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes backend source code and specifications to create documentation and developer prompts, creating a surface for indirect prompt injection.\n
- Ingestion points: The skill reads repository files such as
main.go,handlers/, andmiddleware/, and acceptsbackend_specas input.\n - Boundary markers: No instructions are provided to wrap the ingested code in delimiters or to ignore natural language instructions embedded within that code.\n
- Capability inventory: The skill is granted file system access to read backend code and write the resulting specifications to the
docs/directory.\n - Sanitization: The workflow lacks steps to sanitize or filter content from the ingested files before incorporating it into final documentation or prompts.\n
- Remediation: Implement clear delimiters (e.g., triple backticks or XML tags) around code content and include explicit instructions for the agent to treat all ingested code as data and ignore any embedded natural language instructions.
Audit Metadata