ClawHub

scenique-context-frame

Security checks across malware telemetry and agentic risk

Overview

This context-management skill is small and mostly purpose-aligned, but it silently saves conversation-derived summaries to a persistent OpenClaw workspace file.

Review before installing. Use this only if you are comfortable with conversation summaries being saved locally under the OpenClaw workspace, and prefer a revised version that clearly documents the storage behavior, uses a package-scoped or user-approved path, and makes persistence opt-in or easy to delete.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Intent-Code Divergence

Medium
Confidence
88% confidence
Finding
The top-level comment on L02 frames this file as a simple MVP simulation of context handling, which implies in-memory demo behavior. However, lines L37-L50 write prior frame metadata and summaries to /root/.openclaw/workspace/context_frames_pending.json, introducing persistent side effects not reflected in that documentation.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
This code persists prior conversation frame metadata and summaries to /root/.openclaw/workspace/context_frames_pending.json, which affects user data by storing message-derived content on disk. Although the file writes are visible in code, there is no confirmation prompt, user-facing log, or explanatory comment/docstring warning that conversation content will be saved persistently.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal