registry / @agent-link/server / 0.1.498

@agent-link/server@0.1.498

AgentLink relay server

AI Security Review

scanned 3h ago · by lpm-firewall-ai

LPM treats this as warn-only first-party agent extension lifecycle risk. The package is a relay server with WebSocket agent/client forwarding and controlled local-port tunneling. A user can explicitly install it as an auto-start service. No covert install-time execution or unconsented foreign AI-agent configuration mutation was established.

Static reason

No blocking static signals were detected.

Trigger

User runs `agentlink-server start`, `start --daemon`, or `service install`; an agent and web client then connect to the relay.

Impact

Authorized relay users can access capabilities exposed by a connected agent and its configured local ports; this is a meaningful dual-use remote-control surface.

Mechanism

Explicit daemon/service setup plus encrypted WebSocket relay and local-port proxying.

Rationale

The source establishes a real dual-use remote-agent relay and explicit service persistence, warranting a warning classification under the firewall policy. It does not establish concrete malicious behavior or an unconsented lifecycle attack.

Evidence

package.jsondist/cli.jsdist/service.jsdist/ws-agent.jsdist/ws-client.jsdist/tunnel.jsdist/state-persistence.jsdist/entra.jsdist/index.jsdist/auth-manager.jsdist/session-manager.jsdist/encryption.js

Network endpoints1

/${tenantId}/discovery/v2.0/keys

@agent-link/server@0.1.498

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code