Back to skill

Security audit

Writer Perspective Distiller

Security checks for vulnerabilities and agentic risk

Overview

This is a transparent writing-analysis skill that guides a user through creating a local style-reference SKILL.md, with no hidden execution or data movement found.

Review any generated author-specific SKILL.md before loading it locally, especially if it includes private writing samples, organizational voice guidance, or copyrighted source material. The reviewed package itself is Markdown-only and does not perform automatic installation or execution.

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
  • 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 (3)

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · README.md (reported line 21)May include surrounding context.

md
- Belief candidates must be confirmed by the user — the AI cannot conclude unilaterally
- A blacklist (what she would never write) is mandatory, not just a positive list
- Neighboring authors with surface similarity must be explicitly distinguished
- Before completion, a "test-driven" pass: have the distilled voice rewrite a piece of default-AI prose, and let the user judge whether it really sounds like her

### Output

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

The README instructs the user to place the generated output into ~/.claude/skills/<author-slug>-perspective/SKILL.md, which affects local configuration/state, but it does not explicitly warn about modifying files in the user's environment. For markdown files, SQP-2 applies when behaviors that could affect user data or system integrity are described without a warning.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The document unconditionally includes a dedicated Chinese-version section, which signals a fixed bilingual presentation rather than letting the user choose their preferred language or locale. Under the policy, language constraints should be optional or explicitly justified; no such opt-in or justification is provided here.

Content

No source excerpt is available for this finding.