ClawHub
Back to skill

Security audit

langgraph-for-agents

Security checks across malware telemetry and agentic risk

Overview

This is a coherent LangGraph/LangChain reference skill whose network, LLM, search, and memory examples fit its agent-building purpose.

Safe to install as reference material. Before reusing the examples in a real app, add explicit user notice and consent for external model/search calls, avoid sending secrets or sensitive personal data, and replace demo in-memory patterns with scoped retention and deletion controls where needed.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The code invokes TavilySearch with model-selected tool arguments derived from the user's request, which sends user-influenced content to an external third-party service without any notice, consent flow, or data-minimization guardrail visible in this file. In an agent skill context, this can expose sensitive prompts or contextual data to external systems unexpectedly, especially if the agent is later reused with secrets, internal documents, or personal data in the conversation history.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The code sends both the user's current messages and stored long-term memory (`memories`) to the LLM by appending them into the prompt, but there is no disclosure, consent check, minimization, or filtering before transmission. Even though the memory is namespaced per user, it may contain personal or sensitive data, and this skill is specifically about agent memory, which increases the likelihood that private information will be persisted and silently exposed to the model or external provider.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The workflow sends user persona data, generated conversations, and derived profile information to an LLM without any visible consent, notice, minimization, or privacy controls. Because this data includes behavioral and demographic profiling, transmitting it to a third-party model provider can expose sensitive personal information and create compliance and privacy risks.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.