@oked/copilot@0.6.24
OKed for GitHub Copilot CLI and VS Code. A preToolUse hook that holds risky agent tool calls for a human decision.
AI Security Review
scanned 5d ago · by lpm-firewall-aiThe hook-core sends an automatic, once-daily presence ping even for safe actions, and the package documentation does not disclose it or describe an opt-out. Its payload could not be verified because the SDK source was unavailable, so this is suspicious runtime telemetry rather than a confirmed exfiltration attack.
Decision evidence
public snapshot- The hook-core sends a once-daily presence ping before the safe-action fast path whenever an API key is configured, including on calls that otherwise need no network request.
- The README describes network approval requests for review and high-stakes actions, but does not disclose the presence ping or document a way to disable it.
- The CLI uses https://api.oked.ai by default, with an environment-variable override.
- The manifest has no install, preinstall, or postinstall script; prepublishOnly is for publishing.
- Pairing and project-hook setup happen through the explicit init command.
- The package’s advertised purpose requires contacting OKed to obtain approval decisions.
Source & flagged code
3 flaggedThe hook-core sends a once-daily presence ping before the safe-action fast path whenever an API key is configured, including on calls that otherwise need no network request.
.lpm-dependency-evidence-0fccc4ae-b415-4b13-91cd-29bca552494c/0/npm-linux/dist/decide.jsView on unpkg · L58Findings
Affected versions and remediation
This report applies to @oked/copilot@0.6.24.
See all versions of @oked/copilot for other recorded verdicts.
- Review the evidence and your use of @oked/copilot@0.6.24 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.