@jait/gateway@0.1.887
Jait AI gateway — local-first AI coding agent with terminal, filesystem, and browser control
AI Security Review
scanned 14d ago · by lpm-firewall-aiThe source supports a configurable credential-forwarding risk: an OpenAI key can be sent to an unrestricted custom public endpoint. This is a warning-level dangerous capability, not evidence of a concrete malicious attack.
Decision evidence
public snapshot- Web search obtains the OpenAI key from gateway context or the process environment and accepts a separately configured base URL.
- The OpenAI search request sends that key as a bearer credential to the URL built from the configured base URL.
- The default guard blocks private hosts but does not restrict public destinations to OpenAI.
- Pi MCP provisioning is per-session and is passed to Pi-based providers during launch.
- The package postinstall hook runs a local Graphify runtime provisioning step; this does not establish malicious install behavior.
Source & flagged code
28 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgFindings
Affected versions and remediation
This report applies to @jait/gateway@0.1.887.
See all versions of @jait/gateway for other recorded verdicts.
- Review the evidence and your use of @jait/gateway@0.1.887 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.