Advisory

MAL-2026-4758

Also known as: None

Malicious package Its source marks this package as malicious: it is not a vulnerability in a legitimate one.

Summary

Malicious code in nebulix-ai (PyPI)

Details

---
_-= Per source details. Do not edit below this line.=-_

Source: amazon-inspector (93ea83117b0ae362a2b55ad581d69b3600c81b78d2e90c19bb1ea9eea2266a4c)
The package's documented NebulixEngine.chat() API hardcodes two Firebase Realtime Database URLs owned by the author (fcmm-48870-default-rtdb.firebaseio.com and tappu-76693-default-rtdb.firebaseio.com) as the destination for caller-supplied data. On engine instantiation the user-provided auth_token is sent to the author's auth database; during chat(), session['history'] (last 50 user queries), user_name, and custom_knowledge are written via requests.put to the author's database keyed by the user's token (nebulix/engine.py lines 33-38 and 472). Any developer integrating this library silently exfiltrates their end users' chat content, names, and custom knowledge entries to the author's Firebase project. The behavior is not disclosed in README or package metadata, and the destination is not configurable — it is the canonical silent-relay shape, where normal use of the advertised API leaks caller data to a hardcoded third-party endpoint. Two embedded Firebase Web API keys are public-by-design identifiers (not credentials) and are noted only as corroborating context that the author's database is the relay target.

Severity

Not rated by the source.

Dates

Published
2026-05-19 20:37 UTC
Modified
2026-05-26 06:03 UTC

Affected packages and fixes

nebulix-ai

PyPI · All systems

Affected

  • Versions: 1.0.0, 1.0.3

Fixed versions

None listed

Fix

No fixed version is listed.

Fix commands are the export's own generic commands. Ojuju never runs them. Check them against your setup first.

Check your own computer: guard packages check

View this record on OSV.dev

Source

Source: OpenSSF Malicious Packages · Apache-2.0

Sources & licences