karajan-code@4.49.2
Local multi-agent coding orchestrator with TDD, SonarQube, and code review pipeline
AI Security Review
scanned 10d ago · by lpm-firewall-aiThe postinstall hook performs automatic first-party MCP registration in global agent configuration, which warrants a warning under the agent extension lifecycle policy. The inspected code does not establish a malicious attack chain or support a publish block.
Decision evidence
public snapshot- The package runs scripts/postinstall.js after npm installation.
- The hook writes the package's karajan-mcp entry into the user's Claude configuration and registers its own server in the user's Codex configuration without a consent check.
- The separate ai-trash hook installer requires an explicit install command with the Claude option; the init command asks before installing skills.
- Other flagged locations inspect agent configuration, skip agent files during watching, or define project guidance and ignore rules; they do not establish another install-time agent mutation.
- The automatic setup registers this package's own MCP server and preserves unrelated Claude configuration entries, which fits first-party extension setup.
- The reviewed source shows no network endpoint or credential transfer in this installation flow.
Source & flagged code
24 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgInstall-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgPackage source references child process execution.
scripts/dual-publish.mjsView on unpkg · L17Findings
Affected versions and remediation
This report applies to karajan-code@4.49.2.
See all versions of karajan-code for other recorded verdicts.
- Review the evidence and your use of karajan-code@4.49.2 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.