Context Switch Token Optimizer

• Outcome: Review
• Latest audit: May 29, 2026, 9:26 PM
• Version:
  1.0.1

Overview

The skill is mostly a local context-management tool, but it stores conversation snippets and can change or appear to reset saved context without clear user confirmation.

Review before installing if you plan to use it with confidential conversations. Clear the bundled state file before first use, avoid passing secrets as conversation content, and disable automatic optimization with TOKEN_OPTIMIZER_ENABLED=false if you want context changes to be manual. Use the context_manager.py reset path or manually inspect memory/context_switch_state.json when you need to actually clear saved context.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• 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
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (3)

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The tool prints the first 50 characters of user-supplied conversation content directly to stdout. In a context-management skill, conversations may contain API keys, personal data, credentials, or proprietary project details, so even partial logging can leak sensitive information into terminal history, logs, or orchestration telemetry without user consent.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The compression path mutates in-memory and persisted conversation state by truncating topic history and halving stored core content, then immediately calls save_context_state() without any user confirmation, rollback, or explicit notice. In a context-management skill, this can silently discard user-relevant information and alter future assistant behavior, creating integrity and availability issues for conversation state.

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: The reset path can fully replace the active context state when token usage exceeds thresholds, clearing current topic/history/memory without a user-facing warning or confirmation. Because this component is specifically responsible for preserving and switching conversational context, an automatic reset is more dangerous here: it can unexpectedly erase ongoing work context and cause loss of continuity or user data relied on by downstream interactions.

VirusTotal

64/64 vendors flagged this skill as clean.