@melaya/runner@1.1.76
Run Melaya AI pipelines locally with your own LM Studio or Ollama models
AI Security Review
scanned 5h ago · by lpm-firewall-aiThe flagged credential flows query Anthropic and GitHub for their own provider features, while the CLI requires a user-supplied Melaya token. The package has no install hook; its failure telemetry is bounded and opt-out, and browser downloads require explicit authorization.
Decision evidence
public snapshot- The only lifecycle script is prepublishOnly, which runs the build; there is no install hook.
- The CLI requires a user-provided Melaya runner token and starts provider discovery when the runner is launched.
- Claude credentials are read for model discovery and sent to api.anthropic.com, their own service.
- GitHub Copilot credentials are read for model discovery and sent to api.github.com for a Copilot token exchange.
- Failure reporting is limited to the Melaya service, requires a supplied runner token, and can be disabled with MELAYA_RUNNER_NO_TELEMETRY=1.
- Chromium download requires explicit authorization.
- Provider discovery reads local credential files, but the observed requests use the respective providers’ own API hosts.
- The failure report sends bounded operational fields rather than file contents, paths, usernames, hostnames, or error strings.
- The metadata addresses in the routing hint appear in browser network-blocking policy, not credential collection.
Source & flagged code
20 flaggedPackage source references child process execution.
dist/uvBootstrap.jsView on unpkg · L20Source file is highly similar to a previously finalized malicious package; route for source-aware review.
dist/uvBootstrap.jsView on unpkgFindings
Affected versions and remediation
This report applies to @melaya/runner@1.1.76.
See all versions of @melaya/runner for other recorded verdicts.
- Review the evidence and your use of @melaya/runner@1.1.76 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.