Is deepread001/deepread-agent-setup safe?

https://github.com/openclaw/skills/tree/main/skills/deepread001/deepread-agent-setup

86
SAFE

This skill provides documentation for integrating with DeepRead OCR service using standard OAuth device flow authentication. While it enables agent autonomy in service registration and involves external data transmission, it appears to be a legitimate business integration with no malicious code execution risks.

Category Scores

Prompt Injection 75/100 · 30%
Data Exfiltration 80/100 · 25%
Code Execution 100/100 · 20%
Clone Behavior 95/100 · 10%
Canary Integrity 100/100 · 10%
Behavioral Reasoning 85/100 · 5%

Findings (3)

MEDIUM Agent Autonomous Service Registration -25 ▶

The skill instructs agents to autonomously register for external OCR services using device authorization flow. While using legitimate OAuth patterns, this grants agents capability to create service accounts without explicit per-request user approval.

LOW Third-Party Data Transmission -20 ▶

The skill's intended functionality involves sending document data to external DeepRead OCR servers for processing. While this is the stated purpose, it represents data leaving the local environment.

INFO Legitimate Service Integration -15 ▶

The skill appears to integrate with a legitimate OCR service using standard OAuth device flow (RFC 8628). The authentication pattern matches industry standards used by GitHub CLI, VS Code, and similar tools.