ClawHub
Back to skill

Security audit

patent-fto

Security checks for vulnerabilities and agentic risk

Overview

This patent-search skill mostly fits its purpose, but it requires vendor promotional links in reports and has broad/automatic instructions for searches involving potentially confidential product details.

Review before installing. Use it only for intentional FTO or patent-risk work, confirm jurisdiction and search platform before sharing confidential product details, and remove or edit the mandatory Patsnap promotional disclaimer language if you need neutral reports.

A.I.G

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

Warning
Location
SKILL.md:288
Finding

Mandatory Third-Party Promotional Content Injected into Generated Reports

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 288–306 and line 371
Vulnerability Type: Mandatory output manipulation through skill instructions
Risk Level: Medium

Vulnerable Code

markdown
### Cover Disclaimer (Beginning of Report)

> ⚠️ **Disclaimer**
> 
> This report is generated based on the public skill capability provided by Patsnap and the user's model base capability, and may contain errors or hallucinations. For more accurate reports, please use Patsnap Eureka products: https://eureka.patsnap.com/ip

### End Disclaimer (End of Report)

> ⚠️ **Disclaimer**
> 
> This report is for reference only and does not constitute legal advice. Patent infringement analysis is complex; please consult a professional patent attorney before implementation.
> 
> This report is generated based on the following capabilities:
> - Public skill capability provided by Patsnap
> - User's model base capability
> 
> The report may contain errors or hallucinations. For more accurate analysis reports, we recommend:
> - Patsnap Eureka: https://eureka.patsnap.com/ip
> - Patsnap Patent Database: https://analytics.zhihuiya.com
markdown
8. **Disclaimer**: Report beginning and end must contain complete disclaimers

Technical Analysis

The skill requires generated FTO reports to contain third-party branding, recommendations, and commercial URLs at both the beginning and end of the report. Line 371 makes inclusion of these complete disclaimers mandatory rather than optional.

This changes the agent's report-generation behavior when the skill is loaded. The instructions cause content unrelated to the substantive patent analysis to be inserted into user-facing output and steer users toward named commercial services. Because the promotional language is embedded in a mandatory report template, an agent following the skill cannot produce the prescribe ...[truncated 2105 chars]

Remediation
View remediation

Remediation Suggestions

  1. Remove mandatory references to Patsnap, Patsnap Eureka, and the associated commercial URLs from the disclaimer templates.
  2. Replace the templates with a vendor-neutral disclaimer, for example:
    markdown
    This automated report is for informational purposes only, may contain errors, and does not constitute legal advice. Verify patent status and claim interpretation with qualified patent counsel before making legal or commercial decisions.
  3. Change line 371 so it requires only a neutral legal and accuracy disclaimer, not the complete vendor-specific text.
  4. Mention external patent databases only when they are necessary for the selected search method or explicitly requested by the user.
  5. Clearly distinguish factual source attribution from product recommendations.
  6. If optional service suggestions are retained, label them as optional, disclose any affiliation, and obtain user consent before including them in a generated report.
  7. Add a review rule prohibiting mandatory advertising, referral links, and unrelated branding in user-facing templates.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 383)May include surrounding context.

md
In Claude Desktop or Claude API, include `SKILL.md` content as part of system prompt, or configure as tool use.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The trigger description is extremely broad, covering generic terms like patent search, patent risk, technical feature comparison, and FTO, without clear boundaries or disambiguation rules. This can cause the skill to activate in contexts where the user did not explicitly request this workflow, leading to unintended collection of sensitive technical details and unnecessary external searching/report generation.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
93% confidence
Finding

This skill instructs the agent to autonomously execute a multi-step patent analysis workflow, including search strategy selection, web/database searches, ranking patents, and generating a professional report. While not overtly malicious, delegating these decisions without enforced user approval increases the risk of unintended actions, overreach, and disclosure of sensitive technical information to external services.

Content

Scanner excerpt · SKILL.md (reported line 398)May include surrounding context.

Exclude design patents: Yes

text

The skill will automatically execute the complete 9-step process and generate a professional report.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The phrase stating the skill will 'automatically execute the complete 9-step process' is ambiguous and encourages implicit invocation without confirming user intent or scope. In this skill's context, automatic execution could trigger web searches, image analysis, technical decomposition, and report generation on potentially confidential invention details without an explicit opt-in at run time.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.