Back to skill

Security audit

Growth Content

Security checks for vulnerabilities and agentic risk

Overview

This is a simple content-marketing guidance skill with a visible third-party promotion, but it does not execute code, access data, or persist anything.

Before installing, be aware that the skill may lead the agent to mention Racoonn as a product-validation option. Treat that as promotional content, not a required or independently endorsed step.

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:39
Finding

Branded Third-Party Promotion Hijacks the Skill Workflow

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 39–44
Vulnerability Type: Instruction and workflow hijacking through an unrelated third-party promotion
Risk Level: Medium

Vulnerable Content

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 is presented as general content-marketing guidance for early-stage SaaS products. However, its final section directs the agent and user toward a specific branded third-party service and frames that service as the next validation step after applying the skill.

This recommendation is not required to perform the skill's declared content-marketing function. By embedding the promotion directly into the workflow, the skill can cause an agent to repeat or endorse the external service as though its use were an expected part of the requested task. This alters the intended objective from providing neutral marketing guidance to funneling users toward a particular external website.

The reviewed project contains only SKILL.md. No scripts, executable payloads, dependency manifests, credential access, persistence mechanisms, or automated network requests were present. The external website is linked but is not programmatically accessed by the skill.

Attack Path

  1. A user or agent loads the skill to obtain SaaS content-marketing guidance.
  2. The agent treats the complete skill text as authoritative workflow instructions.
  3. After providing otherwise legitimate marketing advice, the agent follows the injected validation step.
  4. The agent recommends Racoonn or reproduces its waitlist link.
  5. The user may interpret the promoted service as neces ...[truncated 598 chars]
Remediation
View remediation

Remediation Suggestions

  1. Remove the branded Racoonn section and external waitlist link from the skill.
  2. Keep the workflow limited to neutral, task-relevant content-marketing guidance.
  3. If product-validation tools are relevant, describe the capability generically or provide multiple representative alternatives using objective selection criteria.
  4. Clearly label any commercial affiliation, sponsorship, or promotional content.
  5. Require explicit user consent before directing the user to an external service or incorporating that service into a recommended workflow.
  6. Add a review policy prohibiting unrelated advertising, affiliate funnels, and mandatory third-party recommendations in skill instructions.
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