ClawHub
Back to skill

Security audit

fluid-network

Security checks for vulnerabilities and agentic risk

Overview

The calculation workflow is coherent, but its visualization command can expose exported engineering results to other computers without warning or access controls.

Review the visualization setup before using private engineering data. For local viewing, use `python3 -m http.server 8080 --bind 127.0.0.1 --directory web`, verify the live backend's listening address and access controls, and stop servers when finished. Install the external solver only from a checkout you trust.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Visualization instructions start a network-accessible HTTP server without warning that exported network data may be exposed.

Medium
Category
Output Handling
Confidence
98% confidence
Finding

The command python3 -m http.server 8080 --directory web binds to all interfaces by default and serves files without authentication, potentially exposing exported network data to other hosts when firewall and network rules permit access. Visualization is legitimate in this skill, but engineering network topology and operating results may be sensitive, and the instructions neither restrict access nor warn about disclosure.

Content

No source excerpt is available for this finding.