ClawHub
Back to skill

Security audit

Paper Compare

Security checks for vulnerabilities and agentic risk

Overview

This paper-comparison skill is mostly coherent, but it instructs agents to persist detailed research history without clear user consent, retention limits, or deletion guidance.

Review this skill before installing if your paper choices, research goals, or comparison notes could be sensitive. The web/API retrieval behavior fits the tool, but the history feature should ideally be opt-in, minimized, and paired with clear deletion or retention controls.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (4)

External Transmission

Medium
Category
Data Exfiltration
Content
│       └── "Please narrow to 2-5 for meaningful comparison"
    │
    ├── DOI ──→ Fetch via crossref/semantic scholar
    │       └── https://api.crossref.org/works/{doi}
    │
    ├── URL ──→ Fetch via web_fetch
    │       └── Extract title, authors, abstract
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
Use Semantic Scholar API:
```
https://api.semanticscholar.org/graph/v1/paper/{doi}?fields=citationCount
```

### Self-Check: After Retrieval
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill explicitly instructs saving comparison history including user_goal, paper identifiers, topic, comparison angle, verdict, and methodology details to persistent storage, but it provides no user-facing notice, consent step, retention limit, or opt-out. Even though this is research-oriented data rather than obviously secret material, it can still expose sensitive research interests, unpublished work, or proprietary investigation topics if later accessed, leaked, or reused unexpectedly.

Ssd 3

Medium
Confidence
97% confidence
Finding
The history design encourages broad, indefinite retention of structured research context without sensitivity filtering or data minimization. Storing titles, DOIs, URLs, goals, comparison angles, verdicts, and extracted dimensions creates a durable record of user behavior that could reveal confidential research plans, competitive analysis, or regulated subject matter, especially when combined with memory_search for later retrieval.

Static analysis

No suspicious patterns detected.