ClawHub
Back to skill

Security audit

美国投研

Security checks for vulnerabilities and agentic risk

Overview

The skill has a coherent investment-research purpose, but it asks the agent to download or update external code and run it with access to a local Cue API key.

Review this before installing unless you already trust the Cue/Sensedeal runner source. Prefer using a preinstalled, pinned, verified runner, and only proceed if you are comfortable with the agent updating code under ~/.cue, executing it, reading the Cue API key, and spending Cue credits after confirmation.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (3)

Natural-Language Policy Violations

Medium
Confidence
95% confidence
Finding
The manifest description and usage sections are entirely in Chinese and do not indicate that users may interact in other languages or choose their preferred locale. Under the policy, a skill should not impose a specific language unless it offers opt-in or clearly documents a justified region-specific constraint.

Context-Inappropriate Capability

Medium
Confidence
92% confidence
Finding
The skill is presented as a research workflow, but it instructs the agent to bootstrap execution by pulling code from a remote repository and then running a local Python runner. That expands the trust boundary from passive research instructions to arbitrary supply-chain code acquisition and execution, creating clear risk of remote code execution, persistence, or environment compromise if the repo or mirror is malicious or later compromised.

Context-Inappropriate Capability

Medium
Confidence
90% confidence
Finding
The skill instructs use of an API key stored under the user's home directory config, meaning the agent is expected to access local secrets not clearly justified by the manifest alone. Even if intended for legitimate authentication, this behavior increases the chance of unintended secret access, misuse, or leakage to external services when combined with a remotely obtained runner.

Static analysis

No suspicious patterns detected.