ClawHub
Back to skill

Security audit

meta-cognitive-architecture-reconstruct

Security checks for vulnerabilities and agentic risk

Overview

This skill is a local architecture-analysis helper with a small local learning log, not an exfiltration or destructive tool.

Install only if you are comfortable with a local learner file being created or updated. Do not pass secrets, credentials, or private prompt text as learner notes, and review or delete learned_patterns.json if you do not want retained local history.

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
88% confidence
Finding
The skill indicates it can persist data via file writing, but it declares no permissions. Undeclared write capability breaks least-privilege expectations and can mislead operators or orchestrators into granting or invoking behavior they did not review, especially because the skill also describes learner/self-reflection persistence mechanisms.

Tp4

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding
The declared purpose presents the skill as a high-level meta-skill, but the described behavior includes persistent logging to learned_patterns.json, local JSON file processing, architecture analysis, and CLI-style tooling. This mismatch obscures operational capabilities and data handling, which can cause unsafe deployment decisions and increase the chance that file access or persistence occurs without informed consent.

Missing User Warnings

Low
Confidence
92% confidence
Finding
The script persists operational notes to a local JSON file without any notice, consent flow, retention control, or sanitization. Even though this appears intended for telemetry/learning, the stored notes may contain user inputs, prompts, errors, or other sensitive context, creating an unnecessary privacy and data-handling risk if the file is later exposed or reused.

Static analysis

No suspicious patterns detected.