karajan-code@4.46.0
Local multi-agent coding orchestrator with TDD, SonarQube, and code review pipeline
AI Security Review
scanned 12d ago · by lpm-firewall-aiThe postinstall lifecycle hook makes unconsented changes to global Claude and Codex agent configuration, which meets the blocking policy for lifecycle mutation of foreign AI-agent control surfaces. The configured server is package-owned, but that does not make the automatic global configuration change opt-in.
Decision evidence
public snapshot- The package manifest runs scripts/postinstall.js automatically after installation.
- The postinstall script rewrites the user’s global Claude and Codex MCP configuration to register this package’s server, without a consent check.
- The script’s catch handler keeps installation silent on failure; it does not make the configuration change opt-in.
- Other cited locations concern explicit installer or project setup, MCP inventory, configuration checks, or reading project documentation; they do not remove the automatic postinstall behavior.
- The configured MCP server points to this package’s own server, and the Claude setup preserves unrelated environment keys.
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@4.46.0.
See all versions of karajan-code for other recorded verdicts.
- Avoid installing karajan-code@4.46.0. Remove it from direct dependencies and check your lockfile for transitive copies.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
- If this version ran, investigate the affected machine and build environment. Rotate credentials it could access and rebuild from a trusted environment.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.