Security audit

Arquitecto de migracion

Security checks across malware telemetry and agentic risk

Overview

This skill coherently helps plan migrations and generate rollback materials, but users must review generated operational commands before using them on real systems.

Install only if you want migration planning and rollback assistance. Treat generated SQL, cloud, Kubernetes, Terraform, reconciliation, and rollback commands as review drafts: test in staging, keep backups, require explicit approval before production changes, and avoid committing generated reports or runbooks if they contain schema, infrastructure, or contact details.

SkillSpector

By NVIDIA

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 91% confidence
Finding: The skill invokes local Python scripts via shell commands and instructs loading files from the skill directory, which implies effective file-read and shell-execution capabilities, yet it declares no permissions. This mismatch is dangerous because it obscures the skill's actual execution surface from the caller or policy layer, reducing oversight and potentially enabling unintended command, file, or network access through the referenced scripts.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: This guidance includes automated correction logic that can write directly to the target system and overwrite target values from the source, but it does not present explicit safety guardrails, approval requirements, scope restrictions, or rollback expectations adjacent to the examples. In a migration skill, users may copy these patterns into production workflows, causing unintended data modification, propagation of bad source data, or irreversible integrity issues if the reconciliation logic is wrong.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The scheduled workflow describes unattended reconciliation that can auto-correct issues on a recurring basis, but it lacks clear user-facing warnings about the risks of autonomous writes, error amplification, and cascading bad corrections during migrations. Because it normalizes periodic automatic correction in an operational runbook context, users may deploy unsafe self-healing behavior against live systems without sufficient review, safeguards, or kill switches.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The tool writes full compatibility reports to disk, and those reports embed serialized portions of the input schemas plus detailed issues and generated migration scripts. Schema files often contain sensitive internal structure, table names, field names, endpoint definitions, and sometimes defaults or example values; persisting them without an explicit warning or opt-in can cause unintentional data exposure through local files, artifacts, logs, or CI workspace retention.

Missing User Warnings

Low

Confidence: 82% confidence
Finding: The tool writes generated rollback runbooks to disk, and those runbooks include embedded emergency contacts, escalation paths, infrastructure details, and operational recovery procedures. While this is expected functionality, saving such sensitive operational metadata without any warning, redaction option, or output protection increases the chance of accidental disclosure through shared workspaces, logs, backups, or source control.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.