Back to skill

Security audit

Testimonial Builder

Security checks for vulnerabilities and agentic risk

Overview

This is a markdown-only marketing workflow skill for collecting and using testimonials, with privacy and approval considerations called out in its checklist.

Before installing, treat this as marketing guidance rather than legal advice. If you use it with real customers, obtain explicit consent for every displayed identifier, disclose incentives, minimize personal details where possible, and complete legal or compliance review before publishing testimonials in ads, product pages, or creator briefs.

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
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (7)

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 16)May include surrounding context.

md
| **Length** | 2-3 sentences for ads; 4-5 for landing pages; 1-2 for badges | 3-4 sentences for ads; 5-6 for pages | One-word reviews, 300+ word essays |
| **Specificity** | Concrete metrics ("saved $300/month"), specific use cases | Product features mentioned, problem solved | Generic praise ("great!"), no context |
| **Placement** | Above fold on product page, retargeting ads, influencer briefs | Mixed placement, 2-3 channels | Only testimonials page, single channel |
| **Vetting** | Scoring rubric, verify purchase history, require name/photo/video | Basic authenticity check, spot-check grammar | No verification, no quality standards |
| **Format** | Video testimonials (highest converting), photo + quote + name | Text quote with name, verified badge | No identifier, unverified claims |
| **Emotion** | Triumph over objection, surprising result, relief | General satisfaction, feature appreciation | Vague happiness, no emotional hook |

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · assets/testimonial-quality-checklist.md (reported line 139)May include surrounding context.

md
| **Length** | 2-3 sentences for ads; 4-5 for landing pages; 1-2 for badges | 3-4 sentences for ads; 5-6 for pages | One-word reviews, 300+ word essays |
| **Specificity** | Concrete metrics ("saved $300/month"), specific use cases | Product features mentioned, problem solved | Generic praise ("great!"), no context |
| **Placement** | Above fold on product page, retargeting ads, influencer briefs | Mixed placement, 2-3 channels | Only testimonials page, single channel |
| **Vetting** | Scoring rubric, verify purchase history, require name/photo/video | Basic authenticity check, spot-check grammar | No verification, no quality standards |
| **Format** | Video testimonials (highest converting), photo + quote + name | Text quote with name, verified badge | No identifier, unverified claims |
| **Emotion** | Triumph over objection, surprising result, relief | General satisfaction, feature appreciation | Vague happiness, no emotional hook |

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · references/output-template.md (reported line 32)May include surrounding context.

md
| **Length** | 2-3 sentences for ads; 4-5 for landing pages; 1-2 for badges | 3-4 sentences for ads; 5-6 for pages | One-word reviews, 300+ word essays |
| **Specificity** | Concrete metrics ("saved $300/month"), specific use cases | Product features mentioned, problem solved | Generic praise ("great!"), no context |
| **Placement** | Above fold on product page, retargeting ads, influencer briefs | Mixed placement, 2-3 channels | Only testimonials page, single channel |
| **Vetting** | Scoring rubric, verify purchase history, require name/photo/video | Basic authenticity check, spot-check grammar | No verification, no quality standards |
| **Format** | Video testimonials (highest converting), photo + quote + name | Text quote with name, verified badge | No identifier, unverified claims |
| **Emotion** | Triumph over objection, surprising result, relief | General satisfaction, feature appreciation | Vague happiness, no emotional hook |

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The template explicitly collects and publishes personal data such as name, location, title/role, purchase status, and dates, but only mentions usage rights later and does not require privacy notice, data minimization, or explicit consent for public display. In a marketing workflow, this can lead to unauthorized disclosure of customer information, privacy-law noncompliance, and reputational harm if teams treat the template as approval to surface identifying details by default.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

The tone profile for TikTok/Instagram explicitly prescribes "emojis" as part of the expected style, which imposes a specific communication convention for output without offering any user choice or locale/language flexibility. Under the policy, skills should not force a particular language or locale style unless it is optional or clearly justified.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · references/testimonial-scoring-guide.md (reported line 112)May include surrounding context.

md
### Example 3: Generic Review (Low)
**Raw:** "Love this serum! My skin is so much brighter now. Definitely recommend."

- Credibility: 2 (no verification, generic language, no identity)
- Persuasion: 3 (no specificity, no objection handling, no timeline)
- Platform Fit: 2 (could fit many products, no emotional depth)
- **Total: 7 (REJECT)** ✗ Don't use

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
76% confidence
Finding

This markdown guide is entirely written in English and does not indicate any option for alternate language or locale handling. Under the stated policy category, forcing a specific language without user opt-in can be a natural-language policy concern.

Content

No source excerpt is available for this finding.