Skip to content

Security: maic93/agentdock

Security

SECURITY.md

Security Policy

Reporting a vulnerability

Please do not open a public GitHub issue for security vulnerabilities.

Report privately via GitHub's private vulnerability reporting feature on this repository. If that's unavailable to you, email security@agentdock.dev.

Include, as available:

  • A description of the vulnerability and its potential impact.
  • Steps to reproduce (a minimal repro is very helpful).
  • Which package(s) or plugin(s) are affected, and their versions.
  • Whether the issue affects the core kernel/foundation packages, a first-party reference plugin, or the plugin sandboxing model itself.

Response process

  • We aim to acknowledge new reports within 3 business days.
  • We aim to provide an initial assessment (severity, affected versions, planned timeline) within 10 business days.
  • Fixes for confirmed vulnerabilities are released as soon as they are ready and verified; we do not hold fixes for a scheduled release cadence.
  • Once a fix is released, we will coordinate disclosure timing with the reporter and credit them in the advisory unless they prefer otherwise.

Supported versions

Only the latest minor version of each publishable package receives security fixes, in line with the deprecation policy in docs/architecture. Given the project's current pre-alpha status, this policy will be revisited and formalized with version numbers once the first packages are published.

Scope

This policy covers the packages, apps, and first-party plugins in this repository. Vulnerabilities in community/third-party plugins hosted outside this repository should be reported to that plugin's own maintainers; if a vulnerability in a third-party plugin stems from a flaw in the plugin SDK or sandboxing model itself (packages/shared/sdk), please report it here.

Plugin sandboxing disclaimer

AgentDock's plugin sandboxing model is documented in docs/architecture. As with any sandboxing system, we treat sandbox-escape reports as high severity and prioritize them accordingly — please flag these explicitly in your report.

There aren't any published security advisories