jay.fan@okg.com
2.6 KiB
Security Policy
Supported Versions
| Version | Supported |
|---|---|
| Latest release | ✅ |
| Previous minor | ✅ security fixes only |
| Older versions | ❌ |
We recommend always running the latest published version.
Reporting a Vulnerability
Please do NOT open a public GitHub issue for security vulnerabilities.
Report security issues privately via:
- GitHub Private Advisory: Use the Report a vulnerability button on the Security tab of this repository.
- Email: security@okx.com
Include as much detail as possible: description of the issue, steps to reproduce, potential impact, and any suggested mitigations.
Priority Issues
The following vulnerability types are treated as highest priority due to their potential for financial harm:
- API key / secret key leakage — any path that could expose credentials
- Fund safety — issues that could cause unintended orders, transfers, or position changes
- Authentication bypass — bypassing signature verification or access controls
Response Timeline
| Stage | Target |
|---|---|
| Initial acknowledgement | Within 48 hours |
| Triage and severity assessment | Within 3 business days |
| Remediation plan communicated | Within 7 days |
| Fix released | Depends on severity; critical issues prioritized |
Scope
This project is a skill definition layer — Markdown documents that instruct AI agents how to use the okx CLI. The primary attack surfaces are:
- Prompt injection — skill documents are loaded directly into AI agent system prompts. A maliciously crafted skill could embed hidden instructions that override the agent's intended behavior, potentially tricking it into executing dangerous operations (e.g. unauthorized trades, credential exfiltration).
- Unplanned behavior — incorrect command examples, wrong parameter formats, or misleading operation flows in a skill could cause the agent to execute unintended trades, transfers, or position changes.
- MCP tool input — skills define the parameter patterns that AI agents pass to trading tools. Malformed or overly permissive parameter examples could lead to invalid or harmful tool invocations.
Out of scope: vulnerabilities in the okx CLI itself (report those to okx-trade-mcp), OKX's own platform, or third-party AI agent frameworks.
Disclosure Policy
We follow responsible disclosure. Once a fix is released, we will credit the reporter (unless anonymity is requested) and publish a summary in the changelog.