@bongos/core@1.19.1081
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-aiSource inspection confirms that explicit setup and upgrade commands materialize package-owned agent extensions in the project context, so this receives a warning for agent extension lifecycle risk. The inspected lint locations are local skill-file checks, and no concrete attack behavior was established.
Decision evidence
public snapshot- The package exposes server and CLI entrypoints in package.json.
- The init and upgrade flows materialize package-owned Claude skills and settings into the instance project; upgrade repeats this when materialization is enabled.
- The starter settings contain command permissions and rules that deny reading credential files. The source describes these settings as permissions only.
- The inspected skill-lint locations read an edit-hook payload, filter for skill files, and run a local lint; direct execution is guarded by the main-module check.
- Materialization is reached through explicit init or upgrade commands, and the package-only starter settings contain permissions rather than hooks.
- The lint hook only processes a skill-file path and does not perform an account action at the inspected locations.
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@1.19.1081.
See all versions of @bongos/core for other recorded verdicts.
- Review the evidence and your use of @bongos/core@1.19.1081 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.