task-plan-generator_cn

• Outcome: Pass
• Latest audit: May 29, 2026, 8:22 PM
• Version:
  1.0.1

Overview

This planning skill writes local task-planning reports and reads a small set of planning resource files, which is disclosed and aligned with its purpose.

Install only if you are comfortable with local task reports being written under task_reports, possible reads of plan_source.md from the workspace/parent/home resource path, and automatic cleanup of old reports. Avoid using it for sensitive tasks in shared or synced directories unless those reports are acceptable to retain locally.

SkillSpector

By NVIDIA

Vulnerability Patterns

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

Findings (6)

Context-Inappropriate Capability

Medium

• Confidence: 89% confidence
• Finding: The skill recommends shell-based aggregation of archived reports using commands like cat, grep, and jq. In a planning skill, introducing shell execution for routine report parsing expands the attack surface unnecessarily, can process more files than intended via globbing, and may expose sensitive archived task data if executed without strict scoping.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The skill states that it will generate and archive a report after each task, but does not clearly warn the user that task outcomes will be written to disk. This creates undisclosed persistence of task metadata and possibly sensitive execution details, which is a privacy and data-governance issue in a general-purpose planning skill.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill instructs the agent to read plan_source.md from the workspace, parent directory, and the user's home directory without a privacy warning or consent step. Accessing local historical resources outside the immediate task scope can expose unrelated or sensitive information and violates least-privilege expectations.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The skill defines automatic cleanup of archived reports older than 90 days without clearly notifying the user that deletion may occur. Automatic deletion changes user data on disk and can remove audit history or records the user expected to keep, making it unsafe without explicit consent and controls.

Ssd 3

Medium

• Confidence: 95% confidence
• Finding: The skill directs the agent to archive every user confirmation question and answer into task records. User replies often contain file paths, credentials, business context, or other sensitive details, so mandatory verbatim retention creates unnecessary persistent storage of potentially sensitive user-provided data.

Ssd 3

Medium

• Confidence: 97% confidence
• Finding: The skill mandates generating and retaining task briefs for every task result, including tools used, statuses, timing, and historical aggregation. In a general-purpose agent context, this creates a broad activity log that can reveal sensitive workspace operations, user behavior, and operational metadata beyond what is necessary for planning.

VirusTotal

66/66 vendors flagged this skill as clean.