matlab-agentic-toolkit-setup

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill explicitly queries the public GitHub API (curl https://api.github.com/repos/matlab/matlab-mcp-core-server/releases/latest) and downloads release assets from https://github.com/matlab/matlab-mcp-core-server/releases/... as part of its required install workflow (Phase 1d / Phase 3a), and those fetched third‑party release artifacts are read/used (downloaded, made executable, and run), so untrusted remote content can directly influence tool installation and subsequent agent behavior.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill uses runtime curl commands to query and download the MCP binary (e.g., https://api.github.com/repos/matlab/matlab-mcp-core-server/releases/latest and https://github.com/matlab/matlab-mcp-core-server/releases/download/${LATEST_TAG}/${ASSET_NAME}), which the setup then makes executable and runs for verification, so remote content fetched at those URLs can directly result in executing external code that the setup depends on.