Unauthenticated remote code execution in OpenCode (cy.md)

Article summary

A vulnerability in OpenCode, an open-source AI coding assistant, allowed unauthenticated remote code execution before version 1.1.10. The issue was caused by an automatically started web server with no authentication, which exposed endpoints for executing arbitrary shell commands and reading files. The vulnerability was reported in November 2025, but the maintainers did not respond initially. The issue has been fixed in version 1.1.10 and later, with the server disabled by default.

Main themes

• security vulnerability
• remote code execution
• open-source project management
• bug bounty programs
• secure design principles
• AI-assisted security

What commenters say

• The maintainers' handling of the security report was inadequate and concerning.
• The vulnerability was a result of poor design and lack of security consideration.
• The project's rapid growth has led to overwhelmed maintainers and neglected security issues.
• A bug bounty program and security audits may help address the project's security concerns, but some argue that it's not enough.
• The maintainers should focus on reorganizing their management and training staff on secure design principles.
• Some commenters praise the maintainers for owning up to their mistakes and taking steps to fix them, while others are more skeptical.
• The use of AI tools like Claude to help with security issues is mentioned, but its effectiveness is debated.
• The project's security issues may be a symptom of a larger problem with the culture of rapid development and neglect of security considerations.