ClawHub
Back to skill

Security audit

agent-evaluation

Security checks for vulnerabilities and agentic risk

Overview

This skill is a documentation-only agent evaluation guide with a disclosed third-party API example, so the main consideration is avoiding sensitive data in prompts sent to that service.

Before installing, review the external setup guide separately and only provide SKILLBOSS_API_KEY if you intend to use SkillBoss API Hub. Treat prompts, benchmark cases, transcripts, and production traces as potentially sensitive because the example sends message content to a third-party API; redact secrets and regulated data first.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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

Medium
Confidence
92% confidence
Finding
The skill includes example code that sends arbitrary `messages` to a third-party API and authenticates with an environment-sourced API key, but it does not clearly warn users that prompt contents may leave their environment and be disclosed to an external service. In a skill intended for agent evaluation, those messages may contain benchmark data, production traces, or sensitive prompts, making silent transmission materially risky.

External Transmission

Medium
Category
Data Exfiltration
Content
SKILLBOSS_API_KEY = os.environ["SKILLBOSS_API_KEY"]

def call_llm(messages: list) -> str:
    r = requests.post(
        "https://api.heybossai.com/v1/pilot",
        headers={"Authorization": f"Bearer {SKILLBOSS_API_KEY}", "Content-Type": "application/json"},
        json={
Confidence
88% confidence
Finding
This finding is a duplicate manifestation of the same external transmission behavior at the `requests.post` callsite. The risk is not the existence of HTTPS itself, but that the skill offers a ready-made path to transmit potentially sensitive prompts and evaluation artifacts without prominent disclosure or safeguards.

External Transmission

Medium
Category
Data Exfiltration
Content
SKILLBOSS_API_KEY = os.environ["SKILLBOSS_API_KEY"]

def call_llm(messages: list) -> str:
    r = requests.post(
        "https://api.heybossai.com/v1/pilot",
        headers={"Authorization": f"Bearer {SKILLBOSS_API_KEY}", "Content-Type": "application/json"},
        json={
Confidence
88% confidence
Finding
This finding is a duplicate manifestation of the same external transmission behavior at the `requests.post` callsite. The risk is not the existence of HTTPS itself, but that the skill offers a ready-made path to transmit potentially sensitive prompts and evaluation artifacts without prominent disclosure or safeguards.

External Transmission

Medium
Category
Data Exfiltration
Content
def call_llm(messages: list) -> str:
    r = requests.post(
        "https://api.heybossai.com/v1/pilot",
        headers={"Authorization": f"Bearer {SKILLBOSS_API_KEY}", "Content-Type": "application/json"},
        json={
            "type": "chat",
Confidence
85% confidence
Finding
The hardcoded remote endpoint confirms that data is sent to an external service domain outside the user's local environment. In this skill's context, that is significant because evaluation workflows commonly process confidential model behavior data, making unqualified transmission more dangerous than in a generic public-demo example.

Static analysis

No suspicious patterns detected.