@leanandmean/coding-agent@0.74.1-scramjet.53
Coding agent CLI with read, bash, edit, write tools and session management
AI Security Review
scanned 2mo ago · by lpm-firewall-aiThis is a published Scramjet coding-agent library with no registry lifecycle scripts and no hidden exfiltration. Scanner malware-similarity and eval or wasm hits are the product source, Highlight.js, and optional Doom or notify examples, not an attack chain.
Decision evidence
public snapshot- package.json ships only clean, dev, build, copy-assets, and test scripts, with no registry install hooks.
- Shared exec and shell helpers spawn bash or extension commands through child_process as the coding-agent CLI contract.
- Optional example extensions run osascript, PowerShell toasts, or the gh issue list command only after a user loads them.
- The doom-overlay example ships Emscripten glue that loads a local doom.wasm module.
- HTML export vendors Highlight.js v11.9.0 for syntax highlighting.
- dist/index.js only re-exports SDK symbols and does not run a payload on import.
- Dependencies are sibling @leanandmean packages, not a nested self-install of this package with lifecycle hooks.
- This fork removed the pi.dev install telemetry ping; remaining network use is user-run LLM, git, and GitHub tooling.
- Scanner malware-similarity hits are this product's source, a highlight.js vendor file, and optional examples, not a hidden dropper.
Source & flagged code
11 flaggedPackage source references shell execution.
examples/extensions/mac-system-theme.tsView on unpkg · L11Findings
Affected versions and remediation
This report applies to @leanandmean/coding-agent@0.74.1-scramjet.53.
See all versions of @leanandmean/coding-agent for other recorded verdicts.
- Review the evidence and your use of @leanandmean/coding-agent@0.74.1-scramjet.53 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.