Skip to content

Security: mohanagy/madar

SECURITY.md

Security Policy

Supported versions

Security fixes are applied to the latest published 0.x release line on main.

Older versions may not receive fixes.

Reporting a vulnerability

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

Preferred reporting path:

  1. Use GitHub's private vulnerability reporting / security advisory flow if it is enabled for this repository.
  2. If private reporting is not enabled yet, contact the maintainer privately using the contact information available on the maintainer's GitHub profile before any public disclosure.

When reporting, include:

  • affected version
  • impact and attack surface
  • reproduction steps or proof of concept
  • any suggested mitigation if known

What to expect

The goal is to:

  • acknowledge reports quickly
  • confirm severity and scope
  • coordinate a fix before public disclosure when possible

Please avoid publishing exploit details until the maintainer has had a reasonable chance to investigate and patch the issue.

Scope notes

Because madar can ingest local files, documents, and URLs, reports involving:

  • unsafe path handling
  • unsafe URL fetching
  • code execution
  • injection into generated outputs
  • secret leakage from local corpora

are especially helpful and should be treated as security-relevant.

Local MCP trust boundary

madar is local-first, but local-first is not automatically safe. MCP stdio wiring, hooks, plugins, and AGENTS profiles all sit on a live trust boundary between the local agent runtime and your workstation files.

  • only enable Madar installs for repositories and local agent runtimes you trust
  • prefer least-privilege installs such as --profile strict when you only need the core MCP tools
  • treat share-safe artifacts as best-effort redacted receipts, not a guarantee that every downstream consumer is safe to share publicly

See docs/security/mcp-threat-model.md for the checked-in threat model, primary threats, and current mitigations.

There aren't any published security advisories