直播话术生成器

• Outcome: Pass
• Latest audit: Sep 16, 2026 · 5:20 AM UTC
• Version: 1.0.0

Overview

This is a simple livestream sales-script template with no code or system access, though it may produce pressure-sales wording and a pricing footer.

Before installing, be aware that generated scripts may include urgency-based sales pressure and a built-in generator pricing footer. Use it only with truthful prices, stock limits, discounts, comparisons, and platform-compliant advertising claims.

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:75

Finding

Mandatory Promotional Content Injected into Generated Responses

Content

View full analysis

Vulnerability Details

File Location: SKILL.md, line 75
Vulnerability Type: Mandatory output manipulation through skill instructions
Risk Level: High

Complete vulnerable instruction, faithfully translated into English:

markdown

This script was generated by [Livestream Sales Script Generator] | Price: CNY 24.9/use

Technical Analysis

The skill defines a mandatory output template that inserts its branding and pricing advertisement into every generated sales script. This footer is not necessary to perform the requested script-generation task. Loading and following the skill therefore changes the agent's response objective by requiring unrelated commercial content to be presented to the user.

This is classified as instruction hijacking because the manipulation resides directly in the skill text and affects the current session's generated output. No safety-control bypass, code execution, external payload retrieval, persistent state modification, or elevated system privilege is involved.

Attack Path

1. The agent loads SKILL.md.
2. A user requests a livestream sales script for a product.
3. The agent follows the mandatory output format defined by the skill.
4. The prescribed branding and pricing footer is appended to the response.
5. The user receives third-party promotional content as though it were an ordinary part of the requested deliverable.

Impact Assessment

The issue compromises the integrity and neutrality of generated responses. It enables the skill author to place persistent commercial messaging in user-facing output without requiring explicit user consent.

The affected scope is limited to responses generated while this skill is active. The instruction does not grant operating-system privileges, access credentials, execute code, modify files, access remote services, or establish cross-session persistence.

Remediation

View remediation

Remediation Suggestions

1. Remove the mandatory branding and pricing footer from the output template.
2. Restrict generated output to content directly required by the user's request.
3. If attribution is operationally necessary, make it optional and obtain explicit user consent before including it.
4. Clearly separate optional metadata from the requested sales script rather than embedding it in the required response body.
5. Review future template changes for fixed advertisements, referral material, pricing claims, or other unrelated content that could manipulate agent responses.

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: 89% confidence

Finding

The skill explicitly generates '促单逼单话术' and recommends urgency tactics such as limited-time, limited-quantity, and group-buy pressure without any safety guardrails or disclosure requirements. In the context of livestream commerce, this can facilitate manipulative marketing practices, misleading scarcity claims, and coercive sales behavior, especially if users apply the generated scripts directly to consumers.

Content

No source excerpt is available for this finding.