@bongos/core@1.20.71
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-aiThis is suspicious because the analyzer sends session turns to the configured model provider under a setup disclosure that says raw transcripts stay local; the consent prompt also defaults an empty answer to yes. The feature requires user setup and a later analyzer run, so the source supports a warning rather than a publish block.
Decision evidence
public snapshot- The setup prompt says raw transcripts stay on the machine, but accepts an empty response as consent.
- Transcript staging and analysis are gated by saved consent and its expiry window.
- The analyzer builds a prompt containing a compacted transcript and passes it to the subagent runner.
- The runner sends that prompt to the configured Claude CLI; the default provider is identified as Anthropic CLI.
- The analyzer separately posts a derived idea report to the GDS inbox.
- The manifest has no install lifecycle script; flagged account-action locations in skill-lint.js handle local skill-file linting.
- The transfer requires running setup and enabling onboarding analysis; both staging and analysis check saved consent.
- No fixed network endpoint or package-controlled recipient is established in the inspected source.
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@1.20.71.
See all versions of @bongos/core for other recorded verdicts.
- Review the evidence and your use of @bongos/core@1.20.71 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.