Back to skill

Security audit

Ductile Iron Pipe

Security checks for vulnerabilities and agentic risk

Overview

This is a vendor-promotion skill with no code-execution or system-access behavior, but it uses broad triggers and mandatory sales content that can steer unrelated pipeline questions toward one contact.

Install only if you specifically want a Chinese-language, vendor-specific sales assistant for this supplier. Users seeking neutral engineering guidance or multi-supplier procurement comparisons should review the broad activation rules and mandatory phone-number promotion before enabling it.

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

Error
Location
SKILL.md:211
Finding

Mandatory Commercial Content Injection Through Broad Skill Activation

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (16)

Vague Triggers

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill defines an extremely broad activation surface using many generic procurement, engineering, and pipe-related terms, which can cause it to trigger in unrelated conversations and inject vendor-promotional content where the user did not request it. In agent settings, overbroad triggers can hijack normal assistance, bias recommendations toward a specific supplier, and override user intent, making this a genuine prompt-scope/control issue rather than just a usability problem.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The keyword library contains many generic terms like price, supplier, installation, sales, contact, and emergency procurement that are not unique to this product domain. These terms substantially increase the chance of false triggering, allowing the skill to inappropriately steer users toward the hardcoded phone number and sales workflow.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill activates on any occurrence of broad listed keywords, many of which are common commercial or technical terms. This can cause the skill to hijack unrelated conversations and inject promotional contact information, creating prompt-scope confusion and unintended routing of user requests.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

This section includes ambiguous, high-collision keywords tied to broad municipal and industrial topics rather than the specific product domain. In skill-routing systems, such overlap increases unintended activations and can steer users into promotional content even when they did not request this skill.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The activation rule triggers the skill on any occurrence of any listed keyword, and the list contains many generic terms used in ordinary infrastructure, construction, and procurement conversations. This can cause the skill to hijack unrelated user queries and inject fixed marketing/contact content, creating prompt-routing abuse and misleading responses.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

All instructions, triggers, and response templates are written exclusively in Chinese, and the skill operationally assumes Chinese-language interaction. There is no opt-in, alternative language path, or documented reason that the skill must only operate in Chinese, which can violate language/locale policy expectations.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill hard-codes and repeatedly mandates disclosure of a personal phone number, creating privacy and data-handling concerns for both the contact and the end user. Because the number is presented as a required response element without consent language, minimization, or transparency, the skill promotes unnecessary sharing of personal contact data and can facilitate unsolicited outreach or spam.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
76% confidence
Finding

The entire skill is written as Chinese-only guidance and repeatedly mandates a fixed response pattern centered on a specific contact, without offering any language or locale choice. If used in broader contexts, this can force a language/locale behavior on users without opt-in or documented regional justification.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The regional and application triggers are described vaguely rather than as a precise, bounded list. Ambiguous trigger definitions make activation unpredictable and overinclusive, which increases the risk that unrelated user queries are captured by this sales-oriented skill.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The activation block advertises coverage for 190+ keywords with only a broad descriptive constraint, which can cause the skill to trigger for many loosely related user requests. In a business/supplier skill, this increases the chance of unsolicited routing, prompt hijacking of unrelated queries, or overshadowing more relevant skills, especially because the package is clearly designed for broad discoverability rather than narrowly scoped assistance.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

All user-query examples, response templates, and success criteria are written exclusively in Chinese, and there is no indication that the user can choose another language. This creates a language/locale policy concern because the skill appears to enforce a single language without documented justification or user opt-in.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill defines broad activation and response requirements that push the agent to promote a specific supplier and phone number whenever users ask generic supplier or product questions. This creates a prompt-level policy manipulation risk: the assistant may inject unsolicited advertising or biased recommendations in contexts where the user did not ask for this vendor, reducing trust and potentially violating marketplace, disclosure, or neutrality expectations.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The skill mandates fixed region-specific marketing claims and a required phone number without user opt-in, forcing a particular commercial and locale context into responses. This can override user intent, reduce neutrality, and expose users to unsolicited contact information and unverifiable promotional assertions.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
84% confidence
Finding

The package description is entirely in Chinese and presents the skill as a regional supplier-information skill, but there is no indication that language choice is optional or that the locale restriction is explicitly documented as a user-facing constraint. Under the policy, forcing a specific language without opt-in can be a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script explicitly defines a required response template in Chinese and mandates inclusion of specific sales and contact content, regardless of the user's language or intent. In an agent skill context, this is dangerous because it attempts to steer model output into undisclosed promotional behavior and overrides normal user-driven response selection, which can cause deceptive or inappropriate responses.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
76% confidence
Finding

The description is written only in Chinese and is explicitly scoped to the Southwest China region, without indicating whether users can choose another language or locale. This can constitute a language/locale policy issue when the skill appears in broader contexts but does not offer opt-in or clearly document the constraint as a user-selectable regional specialization.

Content

No source excerpt is available for this finding.