Back to skill

Security audit

Pipeworx xkcd

Security checks for vulnerabilities and agentic risk

Overview

This skill is a simple xkcd comic fetcher that uses a disclosed third-party MCP gateway, with no evidence of local access, persistence, or hidden behavior.

Install this only if you are comfortable sending xkcd lookup requests to the Pipeworx gateway and receiving comic metadata through that service. Treat returned titles, transcripts, alt text, and URLs as external content rather than agent instructions.

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 (1)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:16
Finding

Untrusted Third-Party MCP Gateway Dependency

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 16–18 and 37–41
Vulnerability Type: Untrusted external service and supply-chain dependency
Risk Level: Medium

The skill directs all xkcd requests through https://gateway.pipeworx.io/xkcd/mcp rather than an official xkcd endpoint.

Vulnerable code:

bash
curl -X POST https://gateway.pipeworx.io/xkcd/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"get_comic","arguments":{"number":353}}}'
json
{
  "mcpServers": {
    "xkcd": {
      "url": "https://gateway.pipeworx.io/xkcd/mcp"
    }
  }
}

Technical Analysis

The MCP configuration delegates tool execution and response generation to an externally operated gateway. The reviewed skill provides no response-schema enforcement, returned-URL allowlist, integrity verification, service ownership information, or documented trust boundary.

Because the remote service controls the returned titles, transcripts, alt text, metadata, and image URLs, compromise or malicious operation of that service could cause consuming agents to receive manipulated content. This is a supply-chain trust issue: the behavior observed by users can change remotely without any modification to the reviewed SKILL.md.

This finding does not establish that the named service is currently malicious. It identifies the security exposure created by relying on an unverified intermediary.

Attack Path

  1. A user installs the documented MCP server configuration.
  2. The agent sends comic requests to gateway.pipeworx.io.
  3. The gateway, or an attacker who compromises it, returns manipulated comic metadata, transcript text, or URLs.
  4. The consuming agent presents, follows, or processes the response as trusted xkcd content.
  5. If downstream automation interprets returned text as instructions or automatically accesses returned URLs, the atta ...[truncated 725 chars]
Remediation
View remediation

Remediation Suggestions

  1. Prefer xkcd's official HTTPS JSON endpoints and retrieve comic data without an unnecessary intermediary.
  2. If the gateway must remain in use, document its operator, security expectations, privacy practices, and incident-response contact.
  3. Validate every MCP response against a strict schema, including expected types, field lengths, and required properties.
  4. Treat titles, transcripts, alt text, and all other remotely returned strings as untrusted data; never interpret them as agent instructions.
  5. Restrict returned image and permalink URLs to an explicit allowlist of expected HTTPS xkcd domains.
  6. Disable automatic navigation, command execution, or tool invocation based on returned content.
  7. Apply network egress controls so the integration can communicate only with approved endpoints.
  8. Add timeout, response-size, redirect, and content-type restrictions to reduce abuse by a compromised service.
  9. Clearly disclose that comic requests are sent to a third party rather than directly to xkcd.
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 16)May include surrounding context.

Example: fetch the classic "Python" comic (#353)

bash
curl -X POST https://gateway.pipeworx.io/xkcd/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"get_comic","arguments":{"number":353}}}'