thalamus@1.5.87
⚠ Under reviewMobile and Web client for Claude Code
AI Security Review
scanned 18d ago · by lpm-firewall-aiThe inspected runtime telemetry and flagged agent-path helpers do not establish the scanner’s claimed credential or agent-control attack. The postinstall-extracted Linux ARM64 archive remains materially uninspected, so the package cannot be cleared or classified from available source evidence.
Decision evidence
public snapshot- The postinstall hook runs scripts/unpack-tools.cjs, which unpacks a platform archive into tools/unpacked.
- The Linux ARM64 archive expands to over 113 MB, and its trusted offline analysis says the expanded content exceeded the inspection limit.
- Daemon runtime telemetry samples the host name and memory usage, encrypts the samples with an organization key, and posts them to the configured Thalamus server; the daemon starts this telemetry automatically.
- The four flagged agent-configuration locations resolve to helpers that construct Claude project paths.
- The inspected telemetry path sends sampled host hardware details, not credentials or application records, and targets the configured Thalamus server.
- The inspected postinstall script extracts local tool archives; no payload execution or unrelated project mutation was established from the inspected source.
Source & flagged code
20 flaggedPackage defines install-time lifecycle scripts.
package.jsonView on unpkgPackage source references child process execution.
bin/thalamus-mcp.mjsView on unpkg · L2Findings
Affected versions and remediation
This report applies to thalamus@1.5.87.
See all versions of thalamus for other recorded verdicts.
- Review the evidence and your use of thalamus@1.5.87 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.