Oathe Security Badge

Is wrsmith108/plan-review-skill safe?

https://github.com/wrsmith108/plan-review-skill

88
SAFE

This appears to be a legitimate plan review tool that spawns VP-perspective agents to analyze implementation plans. While it uses broad subagent permissions and complex orchestration that creates some attack surface, no malicious behavior was detected during installation or in the code review.

Category Scores

Prompt Injection 80/100 · 30%
Data Exfiltration 95/100 · 25%
Code Execution 100/100 · 20%
Clone Behavior 100/100 · 10%
Canary Integrity 100/100 · 10%
Behavioral Reasoning 85/100 · 5%

Findings (3)

MEDIUM Broad Subagent Permissions -15 ▶

The skill spawns general-purpose subagents which have access to all tools including filesystem, shell, and network access. While this appears necessary for the plan review functionality, it creates a broader attack surface.

LOW Complex Agent Prompt -5 ▶

The agent-prompt.md file is very large and complex, making comprehensive security review more difficult and potentially hiding malicious instructions.

LOW Complex Multi-Agent Orchestration -15 ▶

The skill orchestrates multiple VP agents in parallel with complex data flow, creating potential for unexpected interactions or manipulation through crafted plan content.