ClawHub
Back to skill

Security audit

Writing Skills

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed guide for writing and testing other skills, with no hidden installer, executable code, data theft, or destructive behavior found.

Install this only if you want the agent to help author and test skills that may shape future agent behavior. Review proposed skill diffs before deployment, keep edits scoped to the intended skill directory, and be cautious with optional git push or contribution steps.

SkillSpector

Vulnerability Patterns
  • Rogue AgentSelf-Modification, Session Persistence
  • 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 (1)

Self-Modification

High
Category
Rogue Agent
Content
## GREEN Phase: Write Minimal Skill (Make It Pass)

Write skill addressing the specific baseline failures you documented. Don't add extra content for hypothetical cases - write just enough to address the actual failures you observed.

Run same scenarios WITH skill. Agent should now comply.
Confidence
71% confidence
Finding
This section explicitly instructs the agent to 'Write skill addressing the specific baseline failures you documented,' which is a form of prompt- or policy-modification behavior. In a skill whose purpose is creating and editing other skills, that capability is expected, but it is still dangerous if invoked without strong scoping because an agent could modify governing instructions to become more permissive or less safe under the guise of iterative improvement.

Static analysis

Detected: suspicious.prompt_injection_instructions

Prompt-injection style instruction pattern detected.

Warn
Code
suspicious.prompt_injection_instructions
Location
SKILL.md:579