@lensmcp/vite-plugin@1.25.37
LensMCP Vite plugin — valtio alias, source transform, and browser event bridge.
AI Security Review
scanned 6d ago · by lpm-firewall-aiThe package is an observability plugin and sends replay to its local bridge, but it silently enables session recording by default and disables masking for password and other form inputs. This is undisclosed runtime telemetry with a material risk of capturing sensitive user-entered data, so it warrants a warning rather than a malicious verdict.
Decision evidence
public snapshot- Replay is enabled by default when no option or environment setting is supplied and the capture configuration is absent or unreadable.
- When enabled, the plugin passes replay configuration into the injected browser runtime and constructs its replay endpoint from the bridge host and port.
- The recorder defaults to leaving inputs unmasked and explicitly disables password masking; it sends replay data with HTTP POST.
- The bridge host defaults to 127.0.0.1 and the port is selected locally when unspecified.
- The package describes itself as a frontend observability plugin that streams browser events to LensMCP.
- The plugin normally attaches to Vite's development server, and the inspected replay transport targets the local bridge; no unrelated external collector was identified.
Source & flagged code
8 flaggedPackage metadata claims a different repository identity while copied source loads a runtime dependency bridge.
lib/plugin.jsView on unpkg · L1Replay is enabled by default when no option or environment setting is supplied and the capture configuration is absent or unreadable.
lib/plugin.jsView on unpkg · L1Findings
Affected versions and remediation
This report applies to @lensmcp/vite-plugin@1.25.37.
See all versions of @lensmcp/vite-plugin for other recorded verdicts.
- Review the evidence and your use of @lensmcp/vite-plugin@1.25.37 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.