Bearblog

• Outcome: Review
• Latest audit: May 27, 2026, 8:17 PM
• Version:
  1.0.1

Overview

The skill’s Bear Blog automation purpose is coherent, but its examples include public publishing and deletion flows without enough user confirmation, including bypassing a browser delete confirmation.

Install only if you are comfortable giving the agent access to an authenticated Bear Blog session. Before using it, require explicit confirmation before publish, unpublish, or delete operations, verify the exact post title or URL, and avoid pasting passwords or secrets into ordinary chat unless your environment provides protected secret handling.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (4)

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: The skill explicitly instructs overriding `window.confirm` to force deletion, removing an important safety barrier for a destructive action. In an agent context, this increases the chance of unintended or automated content deletion without meaningful user acknowledgement, especially when combined with browser automation and persisted sessions.

Missing User Warnings

Medium

• Confidence: 83% confidence
• Finding: The publish flow is presented as a simple click with no warning that it makes content publicly visible, and the document also includes unpublish/save operations without emphasizing visibility consequences. In a browser-driven agent, this can lead to accidental publication or visibility changes if the model executes the documented steps without surfacing the risk to the user.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The workflow instructs the agent/user to type an email and password directly into the browser without any warning about sensitive credential handling, secure secret sourcing, or avoiding hardcoded/plaintext credentials in logs and histories. In an agent skill context, this is risky because credentials may be copied into transcripts, retained in memory, or mishandled by automation tooling.

Missing User Warnings

Medium

• Confidence: 84% confidence
• Finding: The workflow includes a direct publish action that can make content publicly visible without documenting any explicit confirmation, review checkpoint, or warning that publication may be irreversible or externally visible. In a browser-automation skill, this increases the chance of accidental publication of drafts, sensitive content, or incorrect posts.

VirusTotal

64/64 vendors flagged this skill as clean.