The manifest has no lifecycle hooks or package entry point, and the flagged behaviors are part of the bundled app’s agent and terminal functions. The reviewed credential handling supports local session resume, while custom OAuth endpoints are explicitly restricted.
Static reason
One or more suspicious static signals were detected.; source matched previously finalized malicious package; routed for review
Trigger
Runtime actions occur when the app or its agent features are used; the package has no install lifecycle hook.
Impact
No unauthorized credential transfer, unrelated install-time execution, or other concrete attack behavior was found.
Mechanism
The reviewed credential access supports local session resume, custom OAuth endpoints are checked against an allowlist, and the flagged shell install command is adapter guidance.
Evidence
~/.claude/.credentials.json~/.centralu
Network endpoints8
api.anthropic.complatform.claude.comclaude.comclaude.aibeacon.claude-ai.staging.ant.dev