v0
Pass
Audited by Gen Agent Trust Hub on Apr 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: Uses
curlandjqto interact with the v0 REST API and process JSON responses. - [EXTERNAL_DOWNLOADS]: Communicates with the well-known service Vercel via its official API domain
api.v0.devto fetch generated code and project metadata. - [DATA_EXFILTRATION]: Follows secure practices by instructing the user to store API credentials in the
V0_TOKENenvironment variable rather than hardcoding secrets. - [PROMPT_INJECTION]: Contains an indirect prompt injection surface as the agent processes natural language output and code generated by an external API.
- Ingestion points: API response body from
https://api.v0.dev/v1/chatsand associated endpoints. - Boundary markers: No specific delimiters or "ignore previous instructions" guards are implemented for the API data.
- Capability inventory: Shell execution (
curl,jq) and file system access (writing to/tmp/v0_request.json). - Sanitization: No explicit sanitization or validation of the remote API content is performed before processing.
Audit Metadata