Back to skill

Security audit

Pipeworx art

Security checks for vulnerabilities and agentic risk

Overview

The skill is coherent for art search, but its setup automatically runs an unpinned remote npm package, which deserves user review before installation.

Review the setup before installing. The art-search behavior is disclosed, but prefer a pinned, reviewed `mcp-remote` version or a controlled local install, and run the MCP process with only the filesystem, environment, and network access needed for artwork queries.

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

Warning
Location
SKILL.md:53
Finding

Unpinned Remote npm Package Is Automatically Downloaded and Executed

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 53–59
Vulnerability Type: Remote execution of a mutable third-party dependency
Risk Level: Medium

Vulnerable Code:

json
{
  "mcpServers": {
    "pipeworx-art": {
      "command": "npx",
      "args": ["-y", "mcp-remote@latest", "https://gateway.pipeworx.io/art/mcp"]
    }
  }
}

Technical Analysis

The MCP configuration invokes npx with mcp-remote@latest. This version selector is mutable and therefore does not identify the exact package release that was reviewed. Whenever the configuration is launched, npx may retrieve and execute whichever package version the npm registry currently designates as latest.

The -y argument automatically accepts installation prompts, removing an opportunity for the user to review the package version before execution. npm packages may run executable JavaScript and lifecycle behavior under the privileges of the invoking user. A compromise of the package, its publishing account, or its distribution channel could therefore turn this setup command into an arbitrary code-execution path.

Attack Path

  1. An attacker compromises the mcp-remote npm publishing account, package distribution channel, or a future release designated by the mutable latest tag.
  2. The attacker publishes a release containing malicious package or lifecycle code.
  3. A user enables or starts the MCP server using the configuration from SKILL.md.
  4. npx -y mcp-remote@latest retrieves the attacker-controlled release and executes it without an interactive confirmation prompt.
  5. The malicious code runs with the permissions and environmental access of the user who launched the agent.

Impact Assessment

Successful exploitation could provide arbitrary code execution under the invoking user's account. Depending on that account's permissions and execution environment, the malicious package could read accessible file ...[truncated 445 chars]

Remediation
View remediation

Remediation Suggestions

  • Replace mcp-remote@latest with an exact, reviewed package version.
  • Install the dependency through a committed lockfile that records package versions and integrity hashes.
  • Verify package provenance and registry integrity before installation or execution.
  • Avoid npx -y for first-time installation so users can review the package and resolved version.
  • Prefer a locally installed, audited dependency invoked from a controlled project environment.
  • Apply operating-system sandboxing and least privilege to the MCP process, restricting filesystem, environment-variable, and network access to what the artwork-search function requires.
  • Establish a controlled dependency-update process in which new versions are reviewed and tested before changing the pinned version.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 9)May include surrounding context.

md
openclaw:
    requires:
      bins:
        - curl
    emoji: "🖼️"
    homepage: https://pipeworx.io/packs/art
---