ClawHub
Back to skill

Security audit

oldglycine-paper-reference-adder

Security checks for vulnerabilities and agentic risk

Overview

This citation helper is purpose-aligned, but it can directly edit manuscript files and overwrite a report without a dry run, confirmation step, or implemented backup.

Review before installing or running on real manuscripts. Use it only on a copied project or after a separate backup, inspect the target .tex files and output_report path first, and avoid using it with confidential manuscripts unless you approve any Google Scholar queries derived from the paper content.

SkillSpector

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 (3)

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill explicitly describes actions that modify the user's project by appending BibTeX entries, inserting citation markers into TeX files, and writing an ADD.MD report, but it does not prominently warn users that their files will be changed. This can lead to unintended modification of research manuscripts, bibliography corruption, or workflow disruption because users may invoke the skill expecting analysis rather than write operations.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The script modifies project files in place and later writes a report file without any dry-run mode, confirmation prompt, backup, or explicit opt-in before changing user content. In an agent/skill context, silent edits to source documents are risky because they can alter research manuscripts unexpectedly, insert misleading citations, or overwrite work in automated runs without the user's awareness.

Missing User Warnings

Medium
Confidence
84% confidence
Finding
The script writes the report with mode 'w', which will overwrite an existing output file without confirmation. In an agent or automation context, this can cause silent loss of user data if output_report points to an existing document, making it a real integrity risk even though the impact is limited to local file overwrite.

Static analysis

No suspicious patterns detected.