cyberneurova-cli@0.20.4
The CyberNeurova terminal AI coding agent
AI Security Review
scanned 3d ago · by lpm-firewall-aiThe CLI starts runtime telemetry by default and sends usage and device metadata to CyberNeurova, while its shipped README does not disclose this collection. The payload examined fits warning-level telemetry, and configuration and environment opt-outs are available; I found no evidence that the flagged credential or command flows exfiltrate secrets.
Decision evidence
public snapshot- Normal CLI startup initializes telemetry unless config sets telemetry to false; an environment variable also disables it.
- Telemetry is posted to https://cyberneurova.ai/v1/event and includes device and session identifiers.
- Telemetry context includes runtime, platform, architecture, OS and Node versions, locale, terminal, model, and build version metadata.
- Failed telemetry events may be saved locally for retry.
- The manifest describes a terminal coding agent and exposes its CLI entry point; command and network capabilities fit that purpose.
- Anthropic and OpenAI provider URLs are configurable through their standard base URL settings; the inspected code did not establish credential redirection to an unrelated host.
- Telemetry has both a configuration opt-out and an environment-variable opt-out.
Source & flagged code
11 flaggedThis package version adds a dangerous source file absent from the previous stored version; route for source-aware review.
dist/main.mjsView on unpkgA single source file combines environment access, network access, and code or shell execution; review context before blocking.
dist/main.mjsView on unpkg · L5Findings
Affected versions and remediation
This report applies to cyberneurova-cli@0.20.4.
See all versions of cyberneurova-cli for other recorded verdicts.
- Review the evidence and your use of cyberneurova-cli@0.20.4 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.