@bongos/core@1.20.58
Cloud Bongos — the AI-first build platform core (GDS + platform surfaces + module system), installed as a versioned dependency (ADR 0108).
AI Security Review
scanned 5d ago · by lpm-firewall-aiThe scanner's install-time concerns are not supported by the manifest. The remaining concern is that an explicit upgrade can overwrite project agent settings, so a warning is appropriate; the inspected behavior does not establish malicious intent.
Decision evidence
public snapshot- The manifest declares no lifecycle hooks, so this behavior is not triggered by package installation.
- The explicit init and upgrade flows materialize project-level Claude skills and settings from the core package; upgrade uses overwrite mode for settings.
- When overwrite mode is used, materialization writes the core settings into the instance settings file, creating a residual risk of replacing project-specific agent configuration during an upgrade.
- The skill-lint hook only reads the touched skill file and prints lint findings; its cited runtime locations do not show account actions.
- Init and upgrade are user-invoked commands, and init's adopt path preserves existing settings.
- The package-owned agent setup and skill linting appear consistent with the product's setup and maintenance functions; no credential transfer or unrelated account action was identified.
Source & flagged code
23 flagged · loading sourcePackage contains a critical-looking secret pattern.
tests/public_mirror_export.mjsView on unpkg · L103AWS access key ID in tests/public_mirror_export.mjs
tests/public_mirror_export.mjsView on unpkg · L103Findings
Affected versions and remediation
This report applies to @bongos/core@1.20.58.
See all versions of @bongos/core for other recorded verdicts.
- Review the evidence and your use of @bongos/core@1.20.58 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.