Back to skill

Security audit

Star Office Deploy

Security checks for vulnerabilities and agentic risk

Overview

This skill is a dashboard deployment guide, but it asks users to run mutable third-party code and can expose local agent data publicly with weak default access controls.

Install only after reviewing or pinning the external Star-Office-UI code, changing the default password and join keys, and deciding whether memory-note display is safe. Avoid public tunneling unless strong access controls are added, and do not enter Gemini keys or prompts that may contain secrets unless third-party processing is acceptable.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (3)

T03 · Remote Payload Retrieval and Execution

Error
Location
SKILL.md:19
Finding

Unpinned Remote Code and Dependency Retrieval

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:76
Finding

Known Default Password and Predictable Shared Join Keys

Content
View full analysis
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Error
Location
SKILL.md:127
Finding

Potential Disclosure of Agent Memory Through Public Dashboard Exposure

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Rogue AgentSelf-Modification, Session Persistence
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (5)

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill instructs the agent to clone and run code directly from an external GitHub repository (git clone, install dependencies, then execute app.py) without pinning a commit, verifying integrity, or reviewing the code. This creates a supply-chain and arbitrary code execution risk on the host, which is broader than simply configuring a local dashboard and could lead to system compromise if the repository or dependencies are malicious or later changed.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
65% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · SKILL.md (reported line 53)May include surrounding context.

或使用后台运行:

bash
nohup python3 app.py > /tmp/star-office.log 2>&1 &

步骤 5: 告知主人访问地址

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The Gemini API setup explains how to send data to a third-party AI service but does not clearly warn that prompts, images, and related content may leave the local environment and be processed by an external provider. For a dashboard that may contain user-generated text or images, this can lead to unintentional disclosure of confidential or personal data.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The Cloudflare Tunnel step exposes the locally hosted service to the public internet, expanding access far beyond local-only deployment. Even if presented as optional, the instructions normalize public exposure without requiring authentication hardening, access restrictions, or confirmation that the user understands the risks, which could allow unauthorized access or discovery of sensitive content.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill describes creating a public URL but does not clearly warn that anyone with the link may reach the service, nor does it discuss privacy, indexing/discovery risk, or the possibility of exposing internal state and agent activity. Omitting these warnings can cause users to unintentionally publish a local dashboard containing sensitive operational or personal information.

Content

No source excerpt is available for this finding.