Back to skill

Security audit

Principled Simplicity Skill

Security checks for vulnerabilities and agentic risk

Overview

ClawScan marks this skill benign with medium confidence: the reported finding is a low-severity concern about prescriptive guidance, while most reviewed security dimensions are rated okay.

The verdict is benign, not a guarantee of safety. Consider the low-severity concern about evaluative guidance, and note that coverage was partial: 9 of 10 components were scanned, the manifest could not be parsed, and the binary image was outside analyzer scope. Install only if the skill’s purpose and its script’s guidance suit your needs.

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
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

The report provides prescriptive architectural guidance and a health benchmark despite explicitly claiming to present observations without diagnosis.

Low
Category
Excessive Agency
Confidence
93% confidence
Finding

The module documentation at L005–L006 reserves diagnosis for Claude, and L157 explicitly says the report presents only observed facts without diagnostic judgments. However, L187–L200 recommends where prompt content belongs (Wiki/Memory, Skill, Code/Validator, or Prompt) and introduces a ≥70% 'healthy' benchmark, which are evaluative judgments rather than objective observations.

Content

No source excerpt is available for this finding.