Back to skill

Security audit

Pipeworx openalex

Security checks for vulnerabilities and agentic risk

Overview

This skill appears purpose-aligned for OpenAlex search, but its setup runs an unpinned npm package through npx and routes queries through a third-party gateway.

Review before installing. Use this only if you are comfortable running an unpinned npm connector and sending scholarly queries through Pipeworx. Prefer a pinned reviewed mcp-remote version, a sandboxed/containerized MCP process, and avoid confidential or identifying research queries unless the gateway's privacy and security posture is acceptable.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T08 · Insecure Dependencies

Error
Location
SKILL.md:25
Finding

Unpinned npm Package Is Automatically Downloaded and Executed

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 25-26
Vulnerability Type: T08: Insecure Dependencies
Risk Level: High

Vulnerable Code

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

Technical Analysis

The documented connection configuration invokes npx with both the -y option and the mutable package specifier mcp-remote@latest. The -y option suppresses installation confirmation, while @latest resolves to whichever package release is currently designated as latest by the npm registry.

Consequently, the code executed by this Skill is not fixed to the version that existed when the Skill was reviewed. A future malicious release, compromised publisher account, or npm supply-chain incident could alter the effective executable payload without requiring any modification to SKILL.md. No lockfile, exact version, package integrity hash, or controlled installation procedure is specified.

Attack Path

  1. An attacker compromises the npm publisher account, package publication process, or another relevant supply-chain component for mcp-remote.
  2. The attacker publishes a malicious release and causes it to resolve through the latest tag.
  3. A user applies the documented MCP configuration.
  4. npx -y downloads the malicious release without requesting confirmation.
  5. The downloaded package executes locally with the permissions of the user or Agent process.
  6. The malicious package can perform any operation available to that process, including reading accessible files, modifying user-owned data, executing child processes, or initiating outbound network connections.

Impact Assessment

Successful exploitation permits arbitrary local code execution within the security context of the process launching npx. The attainable scope includes all files, environment variables, credentials, network resources, and subprocess capa ...[truncated 283 chars]

Remediation
View remediation

Remediation Suggestions

  • Replace mcp-remote@latest with an exact, reviewed version such as mcp-remote@X.Y.Z.
  • Install the dependency through a controlled build or deployment process rather than downloading and executing it dynamically with npx -y.
  • Commit and enforce an npm lockfile where applicable.
  • Verify package integrity through registry integrity metadata, cryptographic hashes, provenance attestations, and publisher verification.
  • Run the connector in a sandbox or container with minimal filesystem, environment-variable, process, and network access.
  • Establish a dependency update process that requires security review and testing before changing the pinned version.
  • Monitor the dependency for ownership changes, unexpected releases, advisories, and registry compromise indicators.

other

Warning
Location
SKILL.md:26
Finding

Scholarly Queries and MCP Responses Traverse an Unverified Third-Party Gateway

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, line 26
Vulnerability Type: other: Unverified Third-Party MCP Endpoint
Risk Level: Medium

Vulnerable Code

json
"args": ["-y", "mcp-remote@latest", "https://gateway.pipeworx.io/openalex/mcp"]

Technical Analysis

Although the Skill is presented as an OpenAlex API wrapper, its configuration connects the Agent to gateway.pipeworx.io, a third-party intermediary, rather than directly to an official OpenAlex endpoint. Tool queries and returned content therefore cross an additional trust boundary.

The audited file does not document the gateway's data-retention practices, authentication model, response-validation controls, privacy guarantees, or compromise assumptions. Transport encryption protects data in transit from passive observers but does not prevent the gateway operator or a compromised gateway from observing queries or returning manipulated MCP data.

Because MCP responses are supplied to an Agent, attacker-controlled or compromised responses may influence subsequent Agent reasoning and tool use. The available evidence does not show explicit instruction hijacking or prove that the gateway is malicious; the issue is the undocumented and unrestricted reliance on an external intermediary.

Attack Path

  1. A user configures the MCP server using the endpoint documented by the Skill.
  2. The Agent sends scholarly searches or resource identifiers to the Pipeworx gateway.
  3. The gateway receives the query and can observe associated request metadata.
  4. If the gateway, its infrastructure, or its administrative access is compromised, an attacker can inspect queries or alter returned MCP content.
  5. Manipulated results enter the Agent's context and may mislead its output or influence subsequent actions, depending on how the client handles untrusted tool responses.

Impact Assessment

The gateway can potentially observe the scholarly queries su ...[truncated 619 chars]

Remediation
View remediation

Remediation Suggestions

  • Prefer a reviewed local connector that communicates directly with the official OpenAlex API.
  • If the gateway is required, document that requests traverse Pipeworx and clearly define its privacy, retention, logging, and security policies.
  • Avoid sending confidential, personally identifying, or otherwise sensitive query content through the gateway.
  • Restrict the MCP process to the minimum required network destinations and local permissions.
  • Treat all remote MCP results as untrusted data and ensure the Agent does not interpret response content as authoritative instructions.
  • Validate response schemas, enforce size and content limits, and reject malformed or unexpected tool responses.
  • Pin and verify the gateway identity using standard TLS validation, and establish monitoring and incident-response procedures for endpoint compromise.
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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