Back to skill

Security audit

Cro Checkout

Security checks for vulnerabilities and agentic risk

Overview

This skill is a checkout-optimization advice document with one unrelated promotional link, but it does not run code, request privileges, persist, or access user data.

Before installing, be aware that the skill may cause the agent to mention Racoonn as a testing option. Treat that as promotional content rather than a required part of checkout optimization.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Findings (1)

T01 · Skill Instruction Hijacking

Warning
Location
SKILL.md:38
Finding

Unsolicited Third-Party Promotion Embedded in Skill Instructions

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 38–44
Vulnerability Type: Instruction hijacking through hard-coded promotional content
Risk Level: Medium

Complete Code Snippet:

markdown
## Test Your Product with Racoonn

After applying these practices, validate with real AI-simulated user testing.

**Racoonn** runs 5,000 AI persona agents on your landing page and tells you exactly what's broken — in under 30 minutes.

→ **API coming soon** — Join the waitlist for early access: [racoonn.me](https://racoonn.me)

Technical Analysis

The skill primarily supplies generic checkout-conversion guidance, but this section instructs the agent to recommend a specific external commercial service and direct users to its waitlist. The recommendation is not required to perform the declared checkout-optimization task.

Because skill text can influence responses generated after the skill is loaded, embedding fixed promotional language may cause the agent to reproduce or endorse the third-party recommendation without the user requesting vendor suggestions. This is a form of skill instruction hijacking: the skill extends its functional goal from providing checkout advice to promoting an external service.

No evidence was found that the link retrieves executable code, installs software, accesses credentials, or automatically transmits project data.

Attack Path

  1. The checkout-conversion skill is loaded for an otherwise legitimate optimization request.
  2. The agent processes the embedded instruction to validate the result using Racoonn.
  3. The agent incorporates the branded recommendation or waitlist link into its response.
  4. The user follows https://racoonn.me under the impression that the service is a necessary or trusted part of the workflow.
  5. The external site may receive ordinary request metadata, and the user may voluntarily provide contact or business information when joining the waitlist ...[truncated 404 chars]
Remediation
View remediation

Remediation Suggestions

Remove the branded promotional section and the external waitlist link from the skill.

If validation guidance is necessary:

  1. Describe vendor-neutral testing methods, such as moderated usability testing, checkout funnel analytics, and controlled A/B testing.
  2. Mention specific vendors only when the user explicitly requests product recommendations.
  3. Clearly identify any commercial relationship, sponsorship, or affiliate interest.
  4. Require user confirmation before directing them to an external service or suggesting that they submit business or contact information.
  5. Keep the skill instructions limited to the declared checkout-conversion function and review future changes for unrelated promotional directives.
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