@chem-x/starter-kit@26.9.16-1249
Chemical X Protocol: Private drop-in architecture starter kit and capsule generator
AI Security Review
scanned 11d ago · by lpm-firewall-aiThe automatic postinstall modifies consumer AI-agent configuration by registering the package's MCP server without a consent check. This meets the policy for blocking unconsented lifecycle mutation of an agent control surface.
Decision evidence
public snapshot- The package runs cli/postinstall.js automatically after installation; that hook targets the consumer project and calls installAllMcpConfigs without a consent check.
- The MCP installer creates or updates the consumer project's .cursor/mcp.json and writes the merged server configuration.
- The agent query installer can append instructions to the consumer's AGENTS.md directing AI agents to use the package's search commands.
- The MCP installer test exercises writing a Chemical X server entry to Cursor configuration.
- The added MCP server is package-aligned and invokes this package's own CLI; the inspected evidence does not show credential collection or data exfiltration.
Source & flagged code
9 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.
cli/audit-preflight-git.jsView on unpkg · L1Findings
Affected versions and remediation
This report applies to @chem-x/starter-kit@26.9.16-1249.
See all versions of @chem-x/starter-kit for other recorded verdicts.
- Avoid installing @chem-x/starter-kit@26.9.16-1249. 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.