@a11ign/screenreader-worker@0.13.8
Drives a real NVDA screen reader through real navigation on Windows and returns what it announced. The HTTP contract is the API.
AI Security Review
scanned 6d ago · by lpm-firewall-aiThe source confirms automatic Windows startup behavior that attempts to disable update and anti-malware protections, creating a concrete, high-impact capability. The package's appliance context leaves intent uncertain, so this warrants a warning rather than a malware block.
Decision evidence
public snapshot- The package exposes the server as its command-line entrypoint.
- Starting the server on Windows calls a trim helper by default unless a skip variable is set or its marker already exists.
- When not elevated, the helper attempts to register and start a scheduled task with highest run level.
- The helper disables services including Windows Update, and its service loop sets each listed service to disabled.
- The helper invokes a command to disable Defender real-time monitoring.
- The package describes its purpose as driving NVDA on Windows, which gives some context for appliance tuning.
- The package describes a screen-reader worker for Windows; the trimming behavior may be intended for a dedicated appliance, but it still changes security settings automatically.
Source & flagged code
7 flagged · loading sourcePackage source references child process execution.
dist/windows-trim.mjsView on unpkg · L1Source spawns a local helper that also contains network and dynamic execution context; review data flow before blocking.
dist/server.mjsFindings
Affected versions and remediation
This report applies to @a11ign/screenreader-worker@0.13.8.
See all versions of @a11ign/screenreader-worker for other recorded verdicts.
- Review the evidence and your use of @a11ign/screenreader-worker@0.13.8 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.