Back to skill

Security audit

OpenQuok Directory Launch

Security checks for vulnerabilities and agentic risk

Overview

The skill provides coherent launch-profile preparation and SEO management, with remote changes disclosed in its full instructions.

This skill can edit OpenQuok profiles and competitor settings and request analytics refreshes; it is not read-only. Specify the workspace, profiles, and competitors you want changed. Review the separately required CLI and openquok-core before providing credentials; directory submissions remain your action.

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
Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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
Findings (4)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The overview and history commands align with reading cached SEO metrics. However, analytics:refresh requests an update, and analytics:competitors:add changes competitor-tracking configuration. These are additional analytics-management capabilities not covered by profile preparation, readiness inspection, or cached-metric reading.

Content

No source excerpt is available for this finding.

The SEO workflow includes remote mutations despite the manifest limiting SEO functionality to reading cached metrics.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

The manifest at L004–L005 describes SEO functionality as reading cached directory SEO metrics. However, L076–L077 direct the agent to invoke a manual SEO refresh and add, remove, or update competitors, which changes remote state rather than merely reading cached data. These operations are also prescribed in the CLI workflow at L056–L057.

Content

No source excerpt is available for this finding.

The command reference exposes SEO competitor-management writes beyond the manifest's cached-metrics reading scope.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

The manifest describes the SEO capability as reading cached directory metrics, but L48–L50 expose commands to add, remove, and update competitor records. These persistent configuration changes are distinct from preparing product profiles or inspecting listing readiness, and L69 also exposes their hosted MCP equivalents.

Content

No source excerpt is available for this finding.

The SEO recipe performs refresh and competitor-write operations despite the manifest limiting SEO functionality to reading cached metrics.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The manifest describes reading cached directory SEO metrics, but L155 invokes directory:analytics:refresh and L156 invokes directory:analytics:competitors:add. These operations refresh analytics and modify competitor configuration rather than merely retrieving cached metrics, extending the declared SEO scope.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.