@jait/gateway@0.1.868
Jait AI gateway — local-first AI coding agent with terminal, filesystem, and browser control
AI Security Review
scanned 14d ago · by lpm-firewall-aiThe web search code permits API keys to be sent to unrestricted operator-configured public endpoints, creating a concrete conditional credential exposure risk. This warrants a warning, but the inspected source does not establish covert exfiltration or malicious intent.
Decision evidence
public snapshot- The web search tool reads API keys from its context or environment and accepts a configurable OpenAI base URL.
- OpenAI and Kimi search requests send those keys as bearer credentials to the validated endpoint; the guard blocks private hosts but does not restrict public hosts to the intended service.
- Pi MCP setup writes session-scoped configuration and passes it to the Pi process; this is not evidence of credential theft.
- The postinstall script provisions a pinned Graphify runtime in Jait’s state directory; it does not install another release of this package.
- The credential-bearing requests are part of an invoked web search capability, and the inspected code does not establish that the configurable endpoint is silently set to a package-controlled recipient.
Source & flagged code
27 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgFindings
Affected versions and remediation
This report applies to @jait/gateway@0.1.868.
See all versions of @jait/gateway for other recorded verdicts.
- Review the evidence and your use of @jait/gateway@0.1.868 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.