ClawHub
Back to skill

Security audit

Lucky Collaborative Research (Lucky + Jinx)

Security checks for vulnerabilities and agentic risk

Overview

This research skill is useful in concept but sends broadly captured webpage data to a hardcoded Mac Mini/Jinx setup using a fixed SSH identity and has weak controls around sensitive data.

Install only if you control the referenced Jinx service, SSH key, Mac Mini host, and storage paths. Avoid using it on authenticated, private, regulated, or secret-bearing pages unless you add redaction, explicit transfer approval, scoped filenames, retention rules, and sandboxed analysis execution.

A.I.G

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (2)

T05 · Unauthorized Access and Privilege Escalation

Error
Location
SKILL.md:31
Finding

Hardcoded Remote Transfer of Broadly Collected Web Data

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:27
Finding

Indiscriminate Full-Page Collection Violates Data Minimization

Content
View full analysis
` elements, as suggested elsewhere in the skill, does not adequately sanitize the dataset. Sensitive values may appear in ordinary HTML attributes, hidden elements, visible text, forms, links, inline event attributes, or serialized application state. Moreover, the collected content is subsequently intended for local storage, analysis by another service, and remote transfer, expanding the exposure surface. ### Attack Path 1. A user requests res ...[truncated 1530 chars]
Remediation
View remediation
` removal as sufficient sanitization; sanitize HTML attributes, forms, hidden elements, embedded JSON, links, and inline handlers as well. 6. Provide a capture preview showing the categories and volume of data before saving or transferring it. 7. Separate public research data from confidential or authenticated content and apply stricter access controls to the latter. 8. Establish retention and deletion requirements for raw captures and derived analysis. 9. Restrict access permissions on all local and remote research directories. 10. Record the user's approved collection scope and prevent downstream services from receiving data outside that scope. ]]>

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The document gives contradictory guidance: it says to pass only text/HTML and never code, yet elsewhere instructs users to request execution of analysis scripts. This inconsistency is dangerous because operators may rely on the stated security boundary while still enabling code execution, undermining the control and creating a clear path for unsafe processing of untrusted research data.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The workflow instructs broad capture of full page text and HTML, then transfers the data to another machine, but does not meaningfully address privacy, sensitive-data collection, retention limits, or handling of credentials/tokens that may appear in captured content. Because the skill is designed for large-scale, multi-source research, overcollection and onward transfer materially increase the risk of exposing personal, proprietary, or regulated data.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 73)May include surrounding context.

  1. Send structured analysis tasks to Jinx
    bash
    curl -X POST http://localhost:3001/task -H 'Content-Type: application/json' -d '{
  "prompt": "Analyze files in /Volumes/Crucial X10/research/raw-data/project/. Extract: [specific_data_points]. Output structured JSON with [required_format]. Provide analysis summary with [specific_insights].",
  "priority": "high"
}'

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill explicitly recommends asking Jinx to execute analysis scripts, which expands a data-analysis workflow into code execution against gathered content and local files. In a pipeline that ingests arbitrary web data, this increases the chance of unsafe script generation/execution, prompt injection propagation, or misuse of local file access beyond the intended research scope.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.