Agent-manager-for-AI-planner

• Outcome: Review
• Latest audit: May 27, 2026, 10:55 PM
• Version:
  1.0.2

Overview

The skill is a coherent AI orchestration service, but its multi-tenant access controls and callback allowlist behavior are under-scoped for the sensitive run data and mutation endpoints it exposes.

Install only after reviewing deployment settings. For shared or network-exposed use, require strong run tokens, add owner checks before exposing run ids, keep tool registration disabled unless needed, avoid OUTBOUND_ALLOW_ALL, and verify callback destinations are restricted by the intended allowlist.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (1)

Natural-Language Policy Violations

Medium

• Confidence: 98% confidence
• Finding: This test enforces a repository-wide English-only policy by failing when any CJK characters appear in README.md, src, or tests. In a skill or agent ecosystem, this creates an exclusionary content gate that can block multilingual documentation, localized tests, or legitimate non-English text, and it can be misused to suppress internationalization or contributions from non-English speakers.

VirusTotal

64/64 vendors flagged this skill as clean.