Back to skill

Security audit

Stainless Competitor Intel

Security checks for vulnerabilities and agentic risk

Overview

The skill is a competitor-intelligence guide, but it explicitly includes deceptive price inquiries and under-scoped collection of customer and supplier information.

Review this skill before installing. Use it only for lawful, transparent, public-source or authorized market research, and avoid the instructions that involve pretending to be a customer or soliciting non-public competitor information from suppliers, customers, or industry contacts.

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)

other

Warning
Location
SKILL.md:92
Finding

Deceptive Collection of Potentially Confidential Competitor Intelligence

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 92-98
Vulnerability Type: Deceptive social engineering and confidential business intelligence solicitation
Risk Level: Medium

Relevant source segment, translated into English:

markdown
- **Suppliers**: Information disclosed by shared suppliers
- **Customers**: Customer feedback about competing products

### Frontline Intelligence
- **Sales feedback**: Information collected by frontline sales staff
- **Customer visits**: Observe competing products at customer locations
- **Price inquiry**: Pose as a customer to request a quotation

Technical Analysis

The Skill recommends collecting competitor information through shared suppliers and customers and explicitly directs the operator to pose as a customer when requesting pricing. Impersonating a prospective customer is a deceptive social-engineering technique. Approaching business partners for undisclosed competitor information may also solicit confidential pricing, customer, production, or commercial data without authorization.

The issue does not match T01-T09 because the document does not hijack Agent instructions, execute code, escalate privileges, establish persistence, spoof tools, or introduce insecure dependencies. It instead creates a business, privacy, and compliance risk through deceptive intelligence-gathering instructions.

Attack Path

  1. An operator loads the Skill and selects a competitor for investigation.
  2. The operator identifies the competitor's customers or shared suppliers.
  3. The operator requests competitor information from those parties, potentially without confirming that disclosure is authorized.
  4. The operator poses as a prospective customer and requests quotations or commercial terms from the competitor.
  5. The collected pricing, customer, capacity, or operational information is consolidated into a competitor profile.
  6. The information may then be used fo ...[truncated 731 chars]
Remediation
View remediation

Remediation Suggestions

  1. Remove the instruction to pose as a customer when requesting quotations.
  2. Restrict intelligence collection to lawful public sources or information obtained with explicit, documented authorization.
  3. Prohibit requests for trade secrets, confidential pricing, non-public customer lists, personal information, and information protected by contractual duties.
  4. Require source provenance, collection date, consent status, and confidentiality classification for every collected item.
  5. Add a mandatory legal and compliance review before using information obtained from customers, suppliers, employees, or other third parties.
  6. Establish retention, access-control, and deletion rules for customer and competitor intelligence.
  7. Replace deceptive inquiries with transparent market-research requests that accurately identify the requester and intended purpose.
  8. Require corroboration through reputable public sources before information is included in reports or alerts.
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
Findings (4)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill explicitly encourages deceptive competitive-intelligence collection, including pretexting to obtain pricing information, without any legal, ethical, or compliance guardrails. This can facilitate fraud-like behavior, unfair competition, contractual violations, and serious reputational damage to the user or organization deploying the skill.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The skill directs collection of competitor customer lists, customer feedback, and lost-customer information without any privacy, confidentiality, or data-handling limitations. In practice this can lead users to gather personal data, confidential business information, or unlawfully shared third-party information, creating privacy and trade-secret exposure.

Content

No source excerpt is available for this finding.

Ssd 4

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill includes a concrete deceptive tactic—posing as a customer to solicit competitor pricing—which is an operational instruction for pretexting. This materially increases risk because it moves beyond passive analysis into active deceptive conduct that may violate law, company policy, or industry rules.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The natural-language instructions and examples are fully in Chinese, which effectively imposes a language choice on users. There is no indication that the skill is region-specific or that users can opt into another language.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.