mistral-deploy-integration

Pass

Audited by Gen Agent Trust Hub on Mar 30, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: Uses platform CLI tools including Vercel, Docker, and gcloud to manage environment variables, create secrets, and deploy application containers.
  • [EXTERNAL_DOWNLOADS]: References the official Mistral AI SDK (@mistralai/mistralai) and well-known container images from Docker Hub (vllm/vllm-openai) and Google Container Registry.
  • [PROMPT_INJECTION]: Identifies an indirect prompt injection surface where user-supplied message data is passed to the Mistral AI API. 1. Ingestion points: req.json() in api/chat.ts extracts messages. 2. Boundary markers: Absent; messages are interpolated directly into the API request. 3. Capability inventory: Uses @mistralai/mistralai to perform chat completions and streaming. 4. Sanitization: Absent; input is passed directly to the model provider as per standard integration patterns.
  • [DATA_EXFILTRATION]: Provides instructions for secure secret management, including adding .env files to .gitignore to prevent credential exposure in version control systems.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 30, 2026, 02:27 PM