Security audit

Iflytek Contract Intelligence Review

Security checks across malware telemetry and agentic risk

Overview

This contract-review skill is purpose-aligned, but users should treat input contracts, configured providers, and generated outputs as confidential.

Install only if you are comfortable reviewing contracts with the OCR, translation, and LLM providers you configure. Use approved private or on-prem endpoints for confidential documents, store outputs in a protected directory, avoid --save-intermediate unless needed, and redact reports before sharing.

SkillSpector

By NVIDIA

Vulnerability Patterns

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

Findings (5)

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The README explicitly describes routing contract contents to OCR, translation, image-understanding, and LLM review services, but it does not warn that sensitive contractual data may be transmitted to external providers. Because contracts often contain confidential business terms, personal data, and regulated information, this omission can cause users to disclose sensitive content without informed consent or appropriate vendor review.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The workflow sends contract content to external OCR, image-understanding, and translation services, and likely to an LLM-backed review client, without any explicit consent gate, warning, or data-handling disclosure in this entrypoint. Because contracts commonly contain highly sensitive commercial and personal information, silent transmission to third-party services can cause confidentiality breaches, regulatory exposure, and unintended retention outside the local trust boundary.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The skill writes cleaned text, clauses, risks, compliance results, and final reports to disk, all of which are derived from potentially confidential contracts, without any explicit warning or consent in the user flow. On shared systems or improperly permissioned environments, this can expose sensitive legal and business data through local persistence long after processing completes.

Missing User Warnings

Medium

Confidence: 87% confidence
Finding: The Markdown report includes raw contract excerpts and risk evidence directly in the generated output, which can expose confidential legal, commercial, or personal data to downstream users, logs, or systems if reports are shared broadly. In a contract-review skill, this is especially sensitive because the input commonly contains proprietary terms, pricing, identities, and signatures, so reproducing source text without disclosure or minimization creates a real privacy and data-handling risk.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The JSON builder serializes broad portions of the full review result, including document metadata, clauses, risks, and other derived content, which can preserve the entire sensitive contract corpus in machine-readable form. Because JSON outputs are easy to store, forward, index, and ingest into other systems, this increases the likelihood and blast radius of unintended disclosure in a legal-document processing context.

VirusTotal

56/56 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.