ClawHub
Back to skill

Security audit

meta-self-play-coevolution

Security checks for vulnerabilities and agentic risk

Overview

The skill is not clearly malicious, but it asks the agent to add persistent self-learning and orchestration behavior without enough limits or user control.

Review this before installing if you do not want a skill that can persist execution notes or encourage self-evolving agent workflows. Avoid passing secrets, private user text, credentials, or sensitive business details into its learner notes unless you are comfortable with local retention and manage deletion yourself.

SkillSpector

Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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)

Lp3

Medium
Category
MCP Least Privilege
Confidence
83% confidence
Finding
The skill advertises no explicit permissions, yet the content indicates memory/reflection behavior such as writing into a learner or self-reflection loop, which implies file or persistent state writes. Undeclared write capability is risky because downstream systems or reviewers may authorize the skill under false assumptions, enabling unintended persistence, tampering, or data retention.

Tp4

High
Category
MCP Tool Poisoning
Confidence
71% confidence
Finding
The skill is described as a meta-skill for self-play/co-evolution, but the body also claims integration into super-agent orchestration, persistent learner injection, and continuous self-evolution. That broader operational scope is security-relevant because it expands autonomy and side effects beyond a narrow transformation skill, making review and policy enforcement harder.

Missing User Warnings

Low
Confidence
92% confidence
Finding
The script persistently records operation outcomes and arbitrary notes to a local JSON file without any consent, disclosure, retention control, or sanitization. Even though this looks intended for telemetry and learning, the `note` field can capture user-provided or sensitive content and store it indefinitely, creating a privacy and data-handling risk.

Static analysis

No suspicious patterns detected.