Mistral OCR

Security checks across malware telemetry and agentic risk

Overview

The skill’s OCR behavior is clearly disclosed, but its open-ended Mistral SDK dependency includes a confirmed compromised release in the allowed range.

Review before installing. The OCR workflow itself is transparent, but install only with a dependency lock or constraint that avoids `mistralai==2.4.6` and verifies the resolved SDK version. Do not use it for sensitive documents unless Mistral cloud processing is acceptable, and prefer a temporary environment variable or secrets manager for the API key.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The script uploads the entire input document to Mistral's remote OCR service, but the user-facing behavior does not clearly and explicitly warn that document contents leave the local machine. This creates a real privacy and data-handling risk, especially if users process sensitive PDFs or images under the assumption that OCR is local.

Unpinned Dependencies

Low
Category
Supply Chain
Content
mistralai>=0.0.9
Confidence
93% confidence
Finding
mistralai>=0.0.9

Known Vulnerable Dependency: mistralai — 2 advisory(ies): GHSA-wx9m-wx4f-4cmg (Malicious dropper in mistralai 2.4.6 PyPI package); MAL-2026-3608 (Malicious code in mistralai (PyPI))

Critical
Category
Supply Chain
Confidence
98% confidence
Finding
mistralai

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal