ClawHub
Back to skill

Security audit

OpenClaw AgentForge Compare

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed remote MCP helper for comparing AI coding tools, with minor routing and documentation scope issues but no malicious or high-impact behavior in the artifact.

Install only if you want an external MCP service to answer AI coding tool comparison, pricing, and recommendation questions. Treat recommendation inputs such as team size, budget, workflow, and must-have features as information sent to the remote service, and verify pricing on official vendor sites before buying.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger phrases are broad enough that the skill may auto-activate for generic recommendation or comparison requests that are only loosely related to AI coding tools. Unintended invocation can cause the agent to route user queries to this external MCP server unnecessarily, creating over-collection of contextual user input and increasing the chance of irrelevant or privacy-impacting tool use.

Vague Triggers

Medium
Confidence
82% confidence
Finding
The profile-oriented triggers include phrases like 'tell me about Cursor' or 'Copilot pricing,' which can overlap with generic product-information requests without confirming that the user wants this specific MCP skill. In an agent ecosystem, ambiguous routing can leak user intent or surrounding context to an unnecessary third-party service and degrade trust through incorrect tool selection.

Intent-Code Divergence

Low
Confidence
96% confidence
Finding
L255 states this server is for 'Compare AI tools, frameworks, MCP servers', yet every declared protocol and tool in the file is limited to AI coding assistants, pricing, categories, and recommendations. This is an active contradiction in the documentation rather than a mere omission, because no tool or input/output contract supports comparing frameworks or MCP servers.

Static analysis

No suspicious patterns detected.