Back to skill

Security audit

Tailscale VPN

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent Tailscale setup guide, but it asks users to run high-impact networking and installer commands without enough safety guidance.

Review this skill carefully before using it on a real workstation or server. Prefer official signed package-manager installation steps over curl-to-shell, run sudo commands only on the intended host, and configure ACLs, subnet routes, exit nodes, and shared access narrowly so you do not unintentionally expose a whole network.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Findings (1)

T03 · Remote Payload Retrieval and Execution

Error
Location
SKILL.md:80
Finding

Unverified Remote Installer Piped Directly to a Shell

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, line 80
Vulnerability Type: Remote payload retrieval and execution
Risk Level: High

Vulnerable Code:

bash
curl -fsSL https://tailscale.com/install.sh | sh

Technical Analysis

The installation instructions download a mutable remote script and immediately pipe its contents to sh. The payload is neither version-pinned nor verified using a cryptographic signature or trusted checksum, and the user has no opportunity to inspect it before execution.

Although the URL uses HTTPS and appears to belong to Tailscale's official domain, those properties do not eliminate the risk. Compromise of the remote endpoint, its distribution infrastructure, DNS or TLS trust chain, or the installer publication process could cause arbitrary attacker-controlled commands to execute locally. The effective code can also change after the Skill itself has been reviewed.

Installing Tailscale is consistent with the Skill's declared VPN functionality. However, piping an unverified remote response directly into a shell exceeds what is necessary to perform that installation safely. Signed operating-system packages or a separately downloaded and verified installer can provide the same functionality with less supply-chain exposure.

Attack Path

  1. An attacker compromises or gains control over the installer endpoint or an upstream component capable of changing its response.
  2. The attacker modifies https://tailscale.com/install.sh to include arbitrary shell commands.
  3. A user or agent follows the documented installation command.
  4. curl retrieves the modified response and passes it directly to sh.
  5. The malicious commands execute with the privileges of the user invoking the command.
  6. If the installer requests elevation or is run from a privileged shell, the payload may gain administrative or root-level access.

Impact Assessment

Successful exploitation permits ...[truncated 608 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove the direct curl | sh pipeline.
  2. Prefer the operating system's package manager and Tailscale's signed package repository, pinning an approved package version where operationally practical.
  3. If a standalone installer is required:
    • Download it to a local file without executing it.
    • Verify a publisher-provided cryptographic signature or a checksum obtained through an independently trusted channel.
    • Review the downloaded script before execution.
    • Execute it only after verification succeeds.
  4. Run installation with the least privileges possible and elevate only the specific package-management or service-management operation that requires it.
  5. Document the expected installer source, version, digest, required privileges, and resulting system changes so future audits can detect unexpected changes.
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (7)

External Script Fetching

High
Category
Supply Chain
Confidence
99% confidence
Finding

Piping a remotely fetched script directly into sh allows immediate code execution from the network without integrity verification, review, or pinning. If the remote source, transport, DNS, or hosting path is compromised, this becomes arbitrary code execution on the target system, often during installation when users are less cautious.

Content

Scanner excerpt · SKILL.md (reported line 80)May include surrounding context.

Linux:

bash
# 一键安装
curl -fsSL https://tailscale.com/install.sh | sh

# 启动服务
sudo tailscaled up

Chaining Abuse

High
Category
Tool Misuse
Confidence
99% confidence
Finding

The '| sh' chain is a classic command-chaining pattern that turns downloaded content into immediate shell execution, removing any opportunity for human review. In this skill's remote-networking context, that is especially risky because users may run it on servers or admin workstations where compromise has broad consequences.

Content

Scanner excerpt · SKILL.md (reported line 80)May include surrounding context.

Linux:

bash
# 一键安装
curl -fsSL https://tailscale.com/install.sh | sh

# 启动服务
sudo tailscaled up

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The listed trigger 安全远程访问 is a generic phrase that could match many ordinary requests about secure remote access, not just Tailscale-specific use. The file does not provide exclusion conditions or tighter activation scope, so invocation boundaries are ambiguous.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill presents network-install and privileged commands as direct copy-paste steps without any warning, verification guidance, or safer alternative. In agent or user workflows, this increases the chance of executing high-risk installation steps on the wrong host or with insufficient review, especially because the instructions include root-level operations and a remote script pipeline.

Content

No source excerpt is available for this finding.

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
92% confidence
Finding

The skill instructs users to run a service-management command with sudo, which normalizes privileged execution without context, validation, or least-privilege guidance. In a skill intended for remote access/networking, encouraging root execution can amplify mistakes or make follow-on compromise more damaging if paired with unsafe install steps.

Content

Scanner excerpt · SKILL.md (reported line 83)May include surrounding context.

curl -fsSL https://tailscale.com/install.sh | sh

启动服务

sudo tailscaled up

text

**iOS/Android:**

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
90% confidence
Finding

Advertising subnet routes with sudo modifies host networking behavior and can expose an entire LAN to the tailnet if misconfigured. In the context of a remote-access skill, this is more dangerous than ordinary admin guidance because it can broaden access boundaries beyond a single device.

Content

Scanner excerpt · SKILL.md (reported line 142)May include surrounding context.

bash
# 假设局域网是 192.168.1.0/24
sudo tailscale up --advertise-routes=192.168.1.0/24

# 在管理后台批准路由
# https://login.tailscale.com/acls

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

The description and operational instructions are presented in Chinese throughout the file, but there is no indication that the skill is intentionally region-specific or that users may choose another language. This can conflict with language/locale policy when a skill effectively forces one language without opt-in.

Content

No source excerpt is available for this finding.