ClawHub
Back to skill

Security audit

agent-learn-external-skill

Security checks for vulnerabilities and agentic risk

Overview

This is a read-only methodology skill with no executable code, credential use, persistence, or data-changing behavior, though its activation wording is broader than ideal.

Install only if you are comfortable with a broad, mostly advisory skill that may activate on generic skill-learning language. Treat its governance and standards claims as author assertions, not as independent security guarantees.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (2)

Vague Triggers

Medium
Confidence
87% confidence
Finding
The trigger keywords are very generic terms like "agent", "learn", and "skill", which can cause the skill to activate in many unrelated conversations. This increases the chance of unintended invocation, context hijacking, or workflow interference, especially in systems that auto-route based on keyword matching rather than explicit user intent.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The manifest description includes many broad trigger phrases such as comparison, benchmarking, improvement, and learning from external skills, which can cause the skill to activate for ordinary user requests beyond its intended scope. Overbroad activation increases the chance of unintended delegation, context capture, and misuse of external-reference workflows in situations where the user did not explicitly request this specialized behavior.

Static analysis

No suspicious patterns detected.