ClawHub
Back to skill

Security audit

OrcaRouter

Security checks for vulnerabilities and agentic risk

Overview

The skill is coherent for using OrcaRouter, but it under-controls remote prompt submission and includes risky CLI install guidance.

Review this skill before installing if you use OrcaRouter with sensitive prompts or billing exposure. Do not let an agent send confidential content through create_chat_completion, create_message, or create_embeddings unless you explicitly intend that data to go to OrcaRouter/OOMOL. Prefer installing the oo CLI through a trusted package or verified instructions rather than blindly running the pipe-to-shell setup command.

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

SkillSpector

Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
Findings (3)

External Script Fetching

High
Category
Supply Chain
Confidence
98% confidence
Finding

The skill recommends curl ... | bash to install software directly from the network, which is a classic supply-chain risk. If the remote host, transport, or script content is compromised, the user may execute arbitrary code immediately with their shell, turning a setup failure into full local code execution.

Content

Scanner excerpt · SKILL.md (reported line 58)May include surrounding context.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger language is excessively broad: 'ANY OrcaRouter request' and 'Whenever a task involves OrcaRouter' can cause the skill to activate for ordinary discussion, analysis, or planning rather than only explicit tool-use requests. In an agent setting, this increases the chance of unnecessary connector execution, unintended data disclosure to the service, and reduced user control over when external actions occur.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The safety section states that untagged actions are reads, but create_chat_completion is an untagged creation action that sends data to a remote API endpoint. Even if it does not mutate persistent user records, treating it as read-only can bypass confirmation and review for data egress, cost-incurring requests, or unintended submission of sensitive prompts to a third party.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.