StruktoLab Studio

AI Security Review

Evidence context: threat category none; evidence quality moderate.

struktolab-studio is a visual editor for .struktolab structogram files, published on OpenVSX by openpatch, with 694 users. It declares no permissions and no host permissions, so it gets no ambient filesystem reach beyond the documents you open and no broad network grant. For a diagram editor, reading and writing the files in your workspace is the job.

The only network finding is NET-FETCH-extension/dist/webview.js-291, a fetch call inside the packaged webview bundle. extension/dist/webview.js is build output from the extension's own toolchain, and a fetch landing there is the webview loading assets or talking to its backend. The endpoints tied to the extension are struktolab.openpatch.org (the publisher's domain), struktolab.new (the shareable link service), and schemas.openxmlformats.org. That last one is the OOXML schema host, which any tool that reads or writes Office document formats needs. A structogram editor that imports or exports documents in those formats has a reason to reach it. None of the three are collection points for source code.

On credentials, there is nothing to report. No finding touches .env, .ssh, .git/config, or cloud credential paths, and nothing references VS Code secret storage or credential providers. The remaining static matches are low-severity code-quality hits on bundled JavaScript: the usual mix of process.env references and Node built-ins that fire on any non-trivial file sitting in a dist/ directory. The IoC entries are the familiar debris of minified code, hex runs and property chains read as hostnames, alongside infrastructure domains such as openxmlformats.org. No malware signature, no obfuscation match, and no tool-poisoning match came back.

The strongest counterargument is the fetch inside a minified webview file that a user cannot read. A webview that calls home could pull remote code or ship document contents out. What argues against that reading: the endpoints are the publisher's own domain and a public schema host, the extension holds no permission to read anything outside the files you open, no secret-reading code exists anywhere in the package, and the fetch sits in first-party build output where webview bundles routinely request assets. A malicious build would need a destination worth sending data to, and there isn't one here.

What a developer is actually installing is a structogram editor that reads and writes your .struktolab files and can reach its publisher's domain plus the OOXML schema host. That matches the described product.

Key Reasons

• No permissions or host permissions are declared, and the network findings resolve to the publisher's own domain (struktolab.openpatch.org), the .new share-link service, and the OOXML schema host (schemas.openxmlformats.org).
• The single network match, NET-FETCH-extension/dist/webview.js-291, sits in bundled build output where webview code legitimately loads assets or contacts its backend.
• No credential-access findings: nothing targets .env, .ssh, .git/config, cloud credential paths, or VS Code secret storage.
• Zero malware-signature, obfuscation, and tool-poisoning matches; the remaining hits are low-severity code-quality rules and IoC fragments from minified JavaScript.
• Publisher and package identity are consistent: openpatch on OpenVSX, 694 users, a structogram editor reaching only the endpoints its own product needs.

False Positive Considerations

• Low-severity code-quality rules firing on bundled JavaScript in extension/dist (process.env references, Node built-ins).
• IoC extractor fragments from minified code, hex substrings and property chains parsed as hostnames, plus known infrastructure domains such as openxmlformats.org.
• A NET-FETCH match inside minified dist/webview.js build output rather than in hand-written extension logic.
• Dependency-category hits generated by npm packages bundled into the webview artifact.

Reviewed 2026-10-12; recommended action: no action; model confidence 82%.