The package is not malicious by install behavior, but it exposes an authenticated remote-code-execution path and default traffic capture in customer pages. This is a dangerous browser-agent capability requiring a warning rather than a publish block on static evidence alone.
Static reason
No blocking static signals were detected.
Trigger
Host application initializes PlunoProductAgent or auto-mounts the widget with a token.
Impact
A compromised or malicious Pluno backend/session could run arbitrary code in the embedding page and access its same-origin browser context; captured application traffic may be disclosed despite redaction.
Mechanism
Remote JavaScript execution plus browser network instrumentation.
Evidence
package.jsondist/product-agent-sdk.jsREADME.mdINTEGRATION.mddist/product-agent-widget.js
Network endpoints3
app.pluno.aiwss://app.pluno.ai/api/product-agent/embed/wsapp.pluno.ai/api/product-agent/embed/attachments/upload