thalamus@1.4.2
Mobile and Web client for Claude Code
AI Security Review
scanned 2mo ago · by lpm-firewall-aiThe package exposes a real remote control surface for Claude MCP commands and an explicit cloud credential-transfer feature. These are risky capabilities, but inspection found no automatic credential exfiltration or foreign AI-agent mutation during installation.
Decision evidence
public snapshot- The install hook extracts bundled archives and marks extracted files executable.
- The CLI defaults to a vendor-controlled remote service and stores local state under the user home directory.
- An explicit connect command obtains Anthropic OAuth tokens and uploads them to the remote service.
- At session startup, remote MCP definitions supply command, arguments, and environment values that are passed into the Claude session.
- An environment-controlled debug mode can send unencrypted logs to a configured remote server.
- The postinstall script is limited to package-local tool extraction and has no network operation.
- OAuth-token upload requires the explicit thalamus connect claude command.
- Remote logging requires two opt-in environment variables.
- No automatic lifecycle hook reads Claude credentials, uploads credentials, or mutates Claude configuration.
- No runtime dependency on another thalamus release was found.
Source & flagged code
10 flaggedPackage defines install-time lifecycle scripts.
package.jsonView on unpkgSource file is highly similar to a previously finalized malicious package; route for source-aware review.
dist/lib.cjsView on unpkgPackage source references dynamic require/import behavior.
dist/lib.cjsView on unpkg · L2Findings
Affected versions and remediation
This report applies to thalamus@1.4.2.
See all versions of thalamus for other recorded verdicts.
- Review the evidence and your use of thalamus@1.4.2 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.