ClawHub
Back to skill

Security audit

opencli-adapter-author

Security checks for vulnerabilities and agentic risk

Overview

The skill appears legitimate for building OpenCLI adapters, but it can inspect browser session data and capture authenticated network responses without enough upfront privacy and authorization guardrails.

Install only for adapter development on sites and accounts you are authorized to test. Use test or low-risk browser profiles where possible, avoid real sensitive sessions, redact cookies/tokens/user data before saving fixtures or logs, and do not replay mutating requests unless you intentionally control the account and impact.

SkillSpector

Vulnerability Patterns
  • 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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (4)

Natural-Language Policy Violations

Medium
Confidence
95% confidence
Finding
The skill explicitly instructs the agent to operate in Chinese without any user opt-in or locale negotiation. This can cause the agent to ignore the user's language preference, leading to misunderstanding of security-relevant instructions, verification output, or remediation guidance in multilingual environments.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The skill explicitly instructs operators to enumerate cookies, localStorage values, and retrieve HttpOnly cookies via page APIs, which can expose authentication material during routine adapter development. In a security-sensitive workflow, normalizing credential inspection without clear handling restrictions increases the risk of accidental credential leakage into logs, prompts, screenshots, or committed code.

Missing User Warnings

Medium
Confidence
87% confidence
Finding
The interceptor guidance tells the operator to perform man-in-the-middle style capture of live page requests and responses, potentially collecting authenticated API payloads and user data, but does not warn about privacy, consent, or minimization. Even if intended for legitimate debugging, this creates a clear path to over-collection of sensitive data and misuse in environments containing real user sessions.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The document gives step-by-step instructions for browser/network reconnaissance, endpoint discovery, token/CSRF investigation, and request replay against third-party sites, but it provides no guardrails about authorization, legal scope, rate limits, privacy, or minimizing impact. In a skill specifically designed to build adapters for external sites, this omission can normalize probing authenticated endpoints and replaying requests in ways that may expose user data, violate site terms, or encourage unsafe collection practices.

Static analysis

No suspicious patterns detected.