@mnemom/mnemom@0.17.27
Transparent AI agent tracing
AI Security Review
scanned 6h ago · by lpm-firewall-aiSource inspection resolves the flagged proxy and child-process behavior as user-invoked gateway routing and certificate validation. The manifest has no install-time hook, and the inspected sources establish no concrete attack behavior.
Decision evidence
public snapshot- package.json exposes CLI binaries and only a prepublishOnly build hook; it has no automatic install hook.
- The reviewed CLI, session, policy, export, skills, identity, and handoff sources describe user-facing operations or an off-by-default handoff tool; the inspected dependency entrypoints are ordinary library exports.
- The Remote Control proxy is started for an explicit --remote-control launch. Its caller sends secrets to the local proxy over stdin and refuses to launch the agent if the proxy does not become ready.
- The proxy handles api.anthropic.com traffic to the configured HTTPS gateway, forwards the Anthropic API key for API-key sessions, and removes the OAuth bearer. This matches the package’s agent-gateway routing behavior.
- The flagged certificate helper invokes openssl to report its version and validate generated certificates; the inspected operation does not send command output over the network.
Source & flagged code
5 flagged · loading sourceA single source file combines environment access, network access, and code or shell execution; review context before blocking.
dist/rc-proxy.mjsView on unpkg · L87Findings
Affected versions and remediation
This report applies to @mnemom/mnemom@0.17.27.
See all versions of @mnemom/mnemom for other recorded verdicts.
- Review the evidence and your use of @mnemom/mnemom@0.17.27 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.