@delorenj/pjangler@1.4.2
Project subsystem bootstrapper CLI
AI Security Review
scanned 1mo ago · by lpm-firewall-aiThere is no npm install-time execution; flagged test files and the new prompt helper are not hidden payloads or credential redirection. The remaining issue is explicit user-command writing of Claude Code hooks, which is dual-use bootstrapper behavior and should warn rather than block.
Decision evidence
public snapshot- User-invoked notebook install writes Claude Code settings by spawning python3 on a packaged projector script.
- The notebook migrate --apply command installs the skill and projects SessionStart and SessionEnd hooks.
- The packaged skill tree under dist/assets/project-notebook-skill is loaded from the CLI and copied as a digest-pinned export.
- The hook fragment registers commands under the user Project Notebook skill path.
- The projector mutates the Claude settings target when its install command runs.
- package.json has no preinstall, install, or postinstall scripts; only prepublishOnly for the publisher.
- tests/test_project_hooks.py is a unittest file shipped in the skill export; the runtime projector is scripts/project-hooks.py.
- dist/prompt.js is a local git and project prompt helper, not a new network or credential path.
- Prompted Telegram and Cloudflare tokens are stored with 1Password after a confirm prompt, not sent to an unofficial gateway.
- Network hosts seen in source are vendor Plane, GitHub, Telegram, Cloudflare, and Trello endpoints.
- Hook projection is skipped unless notebook policy is enabled and the user applies migrate or a matching recipe.
Source & flagged code
5 flagged · loading sourceA package entrypoint or install-reachable source explicitly loads and activates an executable-looking payload from a test, fixture, or hidden path.
dist/index.jsView on unpkgManifest-reachable source sends a prompted API credential to a fixed unofficial gateway and persists the redirection.
dist/index.jsView on unpkgFindings
Affected versions and remediation
This report applies to @delorenj/pjangler@1.4.2.
See all versions of @delorenj/pjangler for other recorded verdicts.
- Review the evidence and your use of @delorenj/pjangler@1.4.2 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.