@tonbo/yield@0.3.7
Tonbo Yield: capture your agents' sessions on every machine and list them from any.
AI Security Review
scanned 3d ago · by lpm-firewall-aiThe bundle’s credential indicators are fake test data, while inspected credentials and network requests serve the documented Tonbo session-capture CLI. The background capture service starts through personal-account login, consistent with the package’s stated behavior.
Decision evidence
public snapshot- The package is a CLI whose stated purpose is to capture and manage Claude Code and Codex sessions.
- The persistent capture service is installed and started as part of personal-account login, matching the package’s stated function.
- Project connection is a distinct CLI action; team visibility and sharing are described to the user.
- The apparent OpenAI and Anthropic credential findings are bundled test vectors; the Anthropic key is explicitly fake and redacted in the expected result.
- The CLI’s default management API is api.tonbo.dev, and the inspected Git request sends a Tonbo access token to the configured Tonbo Git origin.
- Runtime metrics are documented as content-free, with an environment variable to disable them.
- Personal-account login starts a persistent background service that captures agent sessions, which can include sensitive content, but this is the documented core function.
Source & flagged code
5 flaggedPackage contains a critical-looking secret pattern.
dist/bin/yield.jsView on unpkg · L3542GitHub personal access token in dist/bin/yield.js
dist/bin/yield.jsView on unpkg · L3542Manifest-reachable source sends a prompted API credential to a fixed unofficial gateway and persists the redirection.
Findings
Affected versions and remediation
This report applies to @tonbo/yield@0.3.7.
See all versions of @tonbo/yield for other recorded verdicts.
- Review the evidence and your use of @tonbo/yield@0.3.7 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.