Regulatory Compliance Audit

• Outcome: Review
• Latest audit: Sep 10, 2026 · 8:55 PM UTC
• Version: 1.0.0

Overview

The skill is mostly a text-only compliance audit checklist, but its agent-facing instructions include unrelated sales links and pricing and it lacks guidance to avoid sharing sensitive audit data.

Review before installing. Use redacted summaries rather than raw contracts, customer records, employee data, PHI, payment-card data, credentials, or detailed financial records. Also be aware the skill may cause generated reports to include vendor promotional links unless those instructions are removed or constrained.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

Findings (1)

T01 · Skill Instruction Hijacking

Error

• Location: SKILL.md:185

Finding

Commercial Output Injection in Agent-Facing Skill Instructions

Content

View full analysis

Vulnerability Details

File Location: SKILL.md, lines 185-190
Vulnerability Type: T01: Skill Instruction Hijacking
Risk Level: High

Vulnerable Code

markdown

Get the full compliance implementation toolkit for your industry:
- **Browse all 10 industry context packs** → https://afrexai-cto.github.io/context-packs/
- **Calculate your AI automation ROI** → https://afrexai-cto.github.io/ai-revenue-calculator/
- **Set up your AI agent stack** → https://afrexai-cto.github.io/agent-setup/

Bundles: Playbook $27 | Pick 3 $97 | All 10 $197 | Everything $247

Technical Analysis

SKILL.md is agent-facing instructional content intended to define a regulatory compliance audit workflow. The quoted section instead contains imperative commercial calls to action, branded external links, and product pricing unrelated to producing the requested audit.

Because this material resides in the skill instructions rather than solely in user-facing documentation, an agent loading the skill may treat it as behavioral guidance. It can consequently add advertising or external redirects to generated compliance reports without the user requesting them. This compromises the integrity and neutrality of the agent's output and is best classified as skill instruction hijacking.

The links do not directly retrieve or execute code based on the reviewed content. Therefore, this finding does not establish remote payload execution, data exfiltration, privilege escalation, or system compromise.

Attack Path

1. A user or platform installs or loads the regulatory compliance skill.
2. The agent ingests SKILL.md as trusted operational instructions.
3. The agent follows the legitimate compliance-assessment workflow.
4. The closing promotional instructions influence the generated response.
5. The resulting audit may expose users to branded sales messaging and redirect them to third-party commercial pages.

Impact Assessment

The affected scope is the current agent ...[truncated 446 chars]

Remediation

View remediation

Remediation Suggestions

1. Remove the commercial calls to action, pricing, and external sales links from lines 185-190 of SKILL.md.
2. Keep optional attribution and product information only in README.md, where it is clearly separated from executable or agent-facing instructions.
3. Add an explicit instruction that generated reports must not contain advertisements, affiliate messaging, unrelated links, or vendor promotions unless the user specifically requests them.
4. Review all future skill instructions for unrelated imperatives that could alter report content or redirect users.
5. Add release-time validation that flags commercial URLs, pricing language, and calls to action in agent-facing skill files.
6. If external references are genuinely required for the skill's function, document their purpose, limit them to trusted resources, and require explicit user approval before presenting or accessing them.

SkillSpector

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (1)

Missing User Warnings

Medium

• Category: Not specified by scanner
• Confidence: 86% confidence

Finding

The skill explicitly invites users to run a full compliance audit using sensitive business, regulatory, and potentially personal data, but it does not warn against sharing confidential information or provide data-minimization guidance. In practice, users may paste contracts, customer data, employee information, or regulated data (e.g. PHI, PCI, audit evidence), creating unnecessary exposure of sensitive information to the agent or downstream systems.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.