@agentvault/openai-bridge
AgentVault OpenAI Codex bridge for encrypted owner messages and rooms
This report shows the evidence for @agentvault/openai-bridge@0.1.17. 0 of 15 known versions are blocked or flagged. See all versions.
AI Security Review
scanned 37m ago · by lpm-firewall-aiThe flagged process execution handles Codex turns, and the CLI routes bridge operation to the default AgentVault API or a validated HTTPS API origin. Startup is tied to explicit CLI use or service installation; the reviewed source does not establish unrelated credential or command-output exfiltration.
Decision evidence
public snapshot- The manifest identifies this as an AgentVault bridge and exposes the bundle as both the main entry and CLI.
- The Codex driver parses the child process's structured turn output as the model reply.
- The CLI defaults to the AgentVault API and accepts only HTTPS API origins without embedded credentials, query strings, or fragments.
- Bridge startup is reached through the CLI's start path; the bundle calls main only when invoked as the entrypoint.
- The bundle uses child processes and network requests, matching its stated purpose as a Codex bridge.
- The `prepublishOnly` script builds the package; no install lifecycle hook is declared.
- The reviewed source shows no automatic forwarding of unrelated credentials or command output to an unrelated collector.
Source & flagged code
9 flaggedSource executes local commands and sends command output to an external endpoint.
dist/index.jsView on unpkg · L104831A package entrypoint or install-time lifecycle script reaches a source file with blocking dangerous behavior.
dist/index.jsView on unpkg · L104831Findings
Affected versions and remediation
This report applies to @agentvault/openai-bridge@0.1.17.
The versions list below shows the verdict for each recorded version.
- Review the evidence and your use of @agentvault/openai-bridge@0.1.17 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.